In vivo spectroscopy: optical fiber probes for clinical applications

ABSTRACT

Introduction

Fiber optic probe-based in vivo spectroscopy techniques are fast and highly objective methods for intraoperative diagnoses and minimally invasive surgical interventions for all procedures where endoscopic observations are carried out for cancers of different types. The Raman spectral features provide molecular fingerprint-type information and can reveal the subjects’ pathological state in label-free manner, making endoscopy multiplexed fiber optic probe-based devices with the potential for translation from bench to bedside for routine applications.

Areas Covered

This review provides a general overview of different fiber-optic probes for in vivo measurements with emphasis on Raman spectroscopy for biomedical application. Various aspects such as fiber-optic probe, radiation source, detector, and spectrometer for extracting optimum spectral features have also been discussed.

Expert opinion

Optical spectroscopy-based fiber probe systems with ‘Chip-on-Tip’ technology, combined with machine learning, can in the near future, become a complementary diagnostic tool to magnetic resonance imaging (MRI), computed tomography (CT) scan, ultrasound, etc. Hyperspectral imaging and fluorescence-based devices are in the advanced stage of technology readiness level (TRL), and with advances in lasers and miniature spectroscopy systems, probe-based Raman devices are also coming up.

KEYWORDS:

• Optical fiber probe
• cancer diagnosis
• in vivo spectroscopy
• raman spectroscopy
• fluorescence spectroscopy
• point-of-care

1. Introduction

A large part of the financial burden in health care across the world comes from the killer number one, coronary diseases, and second, various cancers. These diseases also cause considerable damage to societal welfare through considerable ‘manpower’ loss and psychological and financial crises for the subjects and people associated with them. Both these remain virtually asymptomatic in the early stages and are diagnosed only when they reach stages where therapy usually is not very effective. Even in developed countries, where regular screening is done for the susceptible population: smokers, females above a certain age, obese candidates, genetically predisposed, etc. The methods used at present for the screening and early detection are mainly direct/endoscopic (colposcopy, gastroscopy, colonoscopy, bronchoscopy, etc.), visual examination, and/or imaging (computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), low-dose computed tomography (LDCT), ultrasound, etc.), giving a preliminary diagnosis, to be further confirmed by biopsy (surgical, fine needle, core needle, endoscopic, cellular lavage, etc.) and pathology [1–3]. These techniques involve the use of costly, complex, non-portable equipment and require highly qualified professionals (radiologists, pathologists, oncologists, etc.), who can do only subjective decision-making based on their experience. The process of screening and early detection often needs several visits to multi-specialty hospitals, which often leads to a huge financial burden. Moreover, these cumbersome processes must be repeated several times during the long periods of therapy may be several months to years. Current screening, diagnostic, and therapy follow-up methods are thus totally inadequate for situations where susceptible individuals have to be screened regularly in large numbers or in pandemic conditions like COVID-19 [4]. This situation can be ameliorated considerably by developing methods and instrumentation which are fast, cost-effective, portable, capable of point-of-care, In Situ, In Vivo operation by trained technicians without the need for highly qualified professionals, and which can give data amenable to statistical analysis with advanced data processing techniques like artificial intelligence (AI)/machine learning (ML) to arrive at objective decision-making without any bias from personal capabilities [5]. Several groups across the world are involved at present in advanced research to develop methods for noninvasive screening, early detection, staging, and follow-up in therapy, especially for the ‘killer’-coronary and various cancer diseases. Optical techniques have been at the forefront of these efforts. These techniques include microscopy, Raman, fluorescence, endoscopy, optical coherent tomography (OCT), optical biopsy, optical imaging, diffuse reflectance (DRS) spectroscopy, etc [6–18]. Literature search using the data bases such as web of Science, PubMed Central, and google scholar has been performed using the keyword ‘fiber optics probe.’ Since the review deals with the origin and evolution of various optical fiber-based spectroscopic techniques for clinical applications with emphasis on Raman spectroscopy technique, there was no time constraint for the literature review. But most of the biomedical applications mentioned are in the past two decades. In this review, the attempt has been made to give more details of the probes so far designed, reported, and their diverse biomedical applications, with focus on cancer detection. The technologies for in vivo fluorescence/diffused reflectance/combination of fluorescence diffused reflectance are matured, hence attention has been given for optical fiber probe development for Raman spectroscopy-based devices.

2. In vivo spectroscopy

Conventional imaging techniques such as CT scan, MRI, PET, and visual endoscopy can give only qualitative information about differences between normal and abnormal conditions for clinical samples such as tissue, cells, and body fluids, and such differences have to be interpreted by the observer subjectively. Vibrational spectroscopy techniques can provide ‘fingerprint’ type information, which is amenable for data processing methods to give an accurate description of the sample and its differences from a normal state, providing information on the type and stage of the disease. With the current state of advances in instrumentation, spectroscopic techniques, and data processing methods, the process of observation and decision-making can be carried out in a few minutes by trained technicians using portable point-of-care (POC) systems [19]. The application of spectroscopic tools for the clinical examination can thus substantially improve the diagnostic ability, monitoring of disease progression, and prognosis of treatment. Mainly three different spectroscopic techniques such as Raman spectroscopy, fluorescence spectroscopy, and diffuse reflectance spectroscopy, in their various forms, are being explored for these applications. Raman and fluorescence spectroscopy techniques have been extensively explored for medical applications such as atherosclerosis detection [20], detection and quantification of biomarkers in body fluids, for the diagnosis of cancers (breast, cervix, and skin), coronary diseases, malaria, and other diseases [17,21–26]. The optical methods can, however, be easily made suitable for in vivo/ in situ, POC observations for screening, observations during surgery, monitoring the efficacy of therapies, etc., by the development of various types of fiber optic probes. This underpins the necessity of designing and developing efficient optical probes which can facilitate the collection of maximum spectral signal from the region of interest, without interference from the signal from other regions/sources, in order to extract the relevant molecular fingerprints. Advances in optical fiber-based probes have received more attention due to their flexibility to access cavities and tubular structures in organs like the lung, esophagus, colon, etc., to look at epithelial surfaces without causing damage like puncture, possible with rigid devices. Researchers have tried a variety of in vivo probes to gain optimal performance from complex tissues within a few minutes. Fabrication and development of disposable probes for clinical applications require careful design because of problems like strong background signals from fibers, reproducible observations at different points during a single examination, etc. Reusable probes need to be fabricated in a way that allows repeated sterilization procedures for routine multiple clinical applications. During the last few years in these research efforts, the spectroscopy community has been witnessing better and better probe configurations dedicated to highly specialized applications [27]. The above-mentioned requirements and potential applications highlight the importance of having a detailed survey of the current developments in optical fiber-based spectroscopic probes.

2.1. Spectroscopy with fiber-Optic probes

The principles and methods of Raman, fluorescence and diffuse reflectance spectroscopy are described in several classic books and reviews [2,28,29]. Raman spectra can provide information on the chemical bonds and groups of the species and act as a molecular fingerprint. The frequency shifts (Raman shift) observed in the scattered radiation provide information on the molecular structure of the constituents of the samples and their physical nature (concentrations, crystal /amorphous phase, symmetry, isotopic composition, etc.), as well as inter-molecular interactions. For in vivo applications multimode fiber is recommended over single mode fiber, for the Raman and the fluorescence study so that it can transmit many modes. For the fluorescence spectroscopy, optical fiber material is silica which has a low attenuation in the low wavelength region e.g. UV-visible range. In the case of Raman, the material of the optic fibers are important, as the fiber fluorescence can mask the Raman scattering signal. The low-OH silica step index fiber are the better suited for the Raman spectroscopy to reduce the OH absorption in the near-IR region. Also the attenuation of this fiber is less compared to high-OH silica fiber. For biomedical application of Raman technique, IR sources are used to avoid the fluorescence from biological sample [30,31]. There are different variants of Raman spectroscopy like resonance Raman scattering(RRS), coherent anti-stokes Raman scattering (CARS), surface enhanced Raman scattering (SERS), Tip enhanced Raman scattering (TERS), and spatially offset Raman scattering (SORS), etc. In addition to the well-established laser-induced fluorescence spectroscopy for in vivo applications, different techniques, like time-resolved fluorescence spectroscopy, confocal fluorescence microspectroscopy, multi-modal spectroscopy, etc. are also available for bio-medical applications [2,18,28,32,33].

3. Experimental systems for fiber-optic spectroscopy

3.1. Conventional laboratory system

The primary requirements to assemble a Raman spectroscopy system include a monochromatic light source, usually a laser for incidence radiation, optics for focusing and collimating, filters to remove unwanted radiation and a sensitive detector to detect the collected scattered radiation. In case of the fluorescence spectroscopy system, the incidence radiation normally uses a tunable source, optics for guiding and spectrometer to detect the emitted radiation. The choices of the laser wavelength, filter and spectrometer-detector systems are highly dependent on the problem under investigation. A schematic diagram of the conventional laboratory experimental setup is shown in Figure 1a.

Figure 1. (a) Conventional laboratory setup for Raman/Fluorescence spectroscopy, in fluorescence spectroscopy notch filter is not required, (b) Schematic diagram of portable fiber-optic-based spectrometer for In Situ/In Vivo applications, (c) Multi-fiber designs: Single incidence multi-collection arrangements:(i).6 collection around-1incidence, (ii) Double-shell collection; 18-around-1, (iii) Triple-shell collection; 36-around-1, adapted and redrawn from [47], (iv) Multi-incidence and multi-collection fiber system, (d) Schematic of a spectroscopy setup with coupled incidence-collection probe head, and (e) Portable laser-induced fluorescence system with the fiber-optic probe mounted on trolley use in the hospital clinic, and (f) Angled probes” and ‘Flat probes.’ Different angled tip probes along with the collection cone angle, adapted and redrawn from [48,49].

The band-pass filter is to pass only monochromatic incidence radiation and focusing lens for focusing of light onto the sample. Collimating collection lens finally collect Raman scattered /fluorescence emission radiation. The notch filter is used to cut off incidence radiation in case of a Raman system. This set up is suitable for both solid and liquid samples under different physical conditions and needs to be slightly modified for in vivo studies.

The experimental setups for fiber-optic probes-based spectroscopy fall into a wide range of configurations depending on the specific problem under investigation. The main objective in such setups is to access the sample in an in situ ‘as-is-where-is’ mode. In general, for such studies, the incidence and spectral recording systems will form one part, and the probe and associated configurational components the second part. The configurations of both parts will depend on the specific problem under investigation. Dedicated clinical applications, like bronchoscopy, colonoscopy, gastroscopy, colposcopy, etc., demand different probe designs since the technical constraints for clinical use may vary from site to site. Optimization of various parameters such as the fiber probe collection efficiency, beam steering properties of the probe, filter designing, probe flexibility, and size are to be considered for clinical applications. In addition, the probe-based optical device needs to be properly enclosed to avoid stray light interference as well as aseptic operation. Further, the device should be compact, miniaturized, and easy-to-use, for point-to-point applications.

3.2. Single-fiber Probe Systems

In the simplest configuration, a single-fiber probe can be used for both incidence radiation and collection of light radiation emanating from the sample. Separation of the two radiations is achieved at the spectrometer level, using appropriate dichroic mirrors and filters, as shown in Figure 1b. All the other optical components are external to the fiber probe, such a system has several advantages. The main units of the system incidence laser, spectrometer, and all required optics can be enclosed in a single portable unit, with only the flexible probe to be handled by the observer. Such a probe can be easily incorporated into conventional endoscopic equipment allowing the adaptation of suitable endoscopes for spectroscopic studies. The major disadvantage of the single fiber probe is that because only one fiber is used for incidence and collection, the signal collected will be only a fraction of the total Raman/Fluorescence signal. Since fluorescence signals are quite strong, such systems are sufficient for fluorescence spectroscopy; while the inherently low efficiency of the Raman scattering process, these probes are not good for the observation of Raman spectra.

3.3. Multi-Fiber Configuration

3.3.1. Flat-Tip Probes

In the second category, separate fibers, or fiber bundles with one or more fibers, are used for incidence and signal collection (Figure 1c). The multi-fiber probe can use higher amounts of radiation spread over larger areas with multi-incidence fibers and can collect much more signal, enabling a much better signal-to-noise ratio in shorter time intervals.

Often in the multi-fiber probes, each fiber bundle starts as separate bundles at their origin, the two bundles are enclosed in a single probe head at the sample end to keep incidence and collection regions identical. The schematics of such a system are shown in Figure 1d. Figure 1e shows the photograph of such a system [11,13] with one end bifurcated (for connection to laser and spectrograph) and the other end with the 6-around-1 configuration (Figure 1c.(i)) enclosed in a single probe head. Here the laser and spectrograph system is mounted on a trolley, and the probe, about 3 meters long, is being used to examine the different sites in the oral cavity. This type of setup has been used to differentiate cervical cancer from chronic cervicitis reported recently [24]. As seen here, the system is quite compact, with both laser and spectrometer in one unit, and the flexible probe will allow measurements at any desired point without the need for moving the entire setup.

Both the single fiber and the multi-fiber probe configurations described here have one major advantage. In both cases, the fiber optic probe has to only transport the incidence radiation to the sample and take the scattered /emission radiation to the spectrometer. All optical parts are outside the probe. There are no optics incorporated in the probe. With present advances in multi-wavelength lasers and light-emitting diodes (LEDs), and mini/echelle spectrometers in this arrangement, one can easily switch between multiple incidence sources, thus, providing the facility of multi-wavelength, multi-modal spectroscopy. This can give additional spectral information, enabling a more precise diagnosis possible [32,34]. In the multi-fiber configuration, one can also use separate fiber probe heads for incidence and signal collection, with one or more fibers being used in each probe. Though this provides some freedom in the choice and placement of the various optical components, the obvious disadvantage is that the two probe heads have to be manipulated together for point-to-point measurements. But it can be used conveniently for biopsy samples which can be translated, while the probe-heads can be kept stationary. The efficiency of the single or multi-probe systems and their versatility will be highly dependent on many other parameters concerning the configuration of the probe. An account of different configurations and their optimization that have been employed by various research groups, followed by the results of their studies, are given in the coming sections.

The design of a fiber probe for Raman application was first reported by Trott and Furtak [35] in which they employed an optic fiber bundle to deliver the laser beam (514.5 nm Argon line, 35 mW power) to the sample, benzene, and another bundle of fibers to collect the scattered radiation. They were quite successful in observing the strong 992 cm⁻¹ C-C symmetric stretching Raman band of benzene. They have also investigated the angle dependence of the efficiency of the collection of Raman scattered radiation for different fiber angles. It was not much effective as most of the scattered radiation was lost due to the fiber cladding. A similar fiber optic bundle was also employed by Schopp et al. in order to transfer Raman-scattered radiation from gases in a laser cavity cell [36].

The use of multiple optical fibers for the collection was demonstrated by McCreery et. al [37] who suggested that the increasing number of collection fibers can enhance the overall efficiency of the probe. Schwab et.al [38] have experimentally demonstrated this using 18 collections optical fibers surrounding a single incidence fiber in a ring-like structure (Figure 1c(ii)). This arrangement significantly increased the collection efficiency of the probe as compared to single optical fiber collection probe, 36 collection fibers around 1 incidence is shown in Figure 1c(iii). The highest fiber probe efficiency is achieved when a thin fiber cladding is used to pack an optimum arrangement of the collection fibers around the incidence optical fiber. When the cladding size is more, the overlap between the light cones of each fiber decreases which adversely affects the probe efficiency [39–42]. The multiple fiber configuration for the collection of the Raman signal is effective up to a certain number of fibers with proper optics to couple with spectrometer so that maximum signal can be collected [43]. But when larger fibers are used, only fewer fibers per ring can be packed efficiently around the incidence fiber to fill the slit of the spectrometer for efficient collection. An overall reduction in probe efficiency was observed in these types of designs reported in the literature [44,45]. In many instances, like screening for infectious diseases [16], monitoring hazardous samples like viral and bacterial specimens, or samples at a distance (e.g. operation theater (OT) where only the fiber needs to be in the OT, the rest of the instrumentation kept out of OT), the above probes may not be sufficient, especially when detection of smaller concentrations of components are required. Often such samples may have a uniform composition, (e.g. clinical samples like saliva and blood, environmental samples like waste water, effluents, etc.) and larger volumes/areas can be studied with higher powers at the same time. For such applications, the fiber probe with the multiple incidence -collection configurations (Figure 1c (iv)) can be advantageously used.

3.3.2. Angled probes

Several modifications of the configurations in Figure 1 c have been tried. For example, the fiber probes can also be of two types, probes where the tips of the fibers are flat (incidence and collection fibers both are flat, as discussed above), or the collection probes can be angled. It has been observed that the collection efficiency can be improved with such a configuration because of the larger areas covered by the collection probes in the angled probe configuration. Designs for such ‘Different-Angled’ fiber probe head designs are shown in Figure 1 f for comparison with the traditional flat face configuration. In the angled probe designs, the incidence and collection fibers can either be placed at an angle to each other, or the tips of the collection fibers can be beveled at an angle. In other words, the collection fibers are angled inwards. i.e. toward the central light delivery fiber [46] (Figure 1 f(v)), or both the collection and incidence fibers are kept normal to the imaging plane, and the collection-fiber tips are cut at a certain angle to the normal [47] (Figure 1 f(vi)). Both of these approaches are complex to fabricate compared to the flat fiber probe. In the probe configuration developed by McLachlan et al [48] (Figure 1 f(v)), the collection fibers were physically angled or bent toward the incidence fiber optical axis in such a way that it could facilitate complete overlap of the light cone between the incidence and collection cone. The bending angle was optimized in between the range 0 ° to 45 ° to provide more efficiency. To maintain the required convergence angle, either the fibers were epoxied together, or shims were placed between them and the probes were enclosed inside a protective metal sheath to ensure proper angle between collection fibers and incidence n fiber is maintained in addition to supporting and protection. Even though the overlap cone obtained will be maximum in the case of Figure 1 f(v), fabricating this type of fiber probe is difficult compared to the flat tip fiber probe.

O’Rourke and Livingston [49] developed and patented single delivery and single collection fiber probe (Figure 1 f(iii)), normally named the chisel-tip design. The beveled angle is below 20°. This configuration enhanced the coupling efficiency relative to the flat face single delivery and collection probes (Figure 1 f(i)). For a silica-air interface, this method of angled end faces toward each other makes both light cones to be direct toward each other, facilitating a better overlapping resulting in higher efficiency [50,51] (Figure 1 f(ii)). This probe design is claimed to be approximately five times more effective than traditional designs in the perspective of light coupling. Another variant of this probe design reported involves the use of 6 fibers around one incidence fiber [52], where the faces of all the collection fibers are beveled (Figure 1 f(vi)). This version of the probe was further improvised and obtained an approximately fivefold increase in signal level via the efficient coupling of the light cones compared to the ‘6 around 1’ flat-fiber design used by O’Rourke and Livingston [49].

4. Probes with other configurations

In the multi-probe systems described above, one can incorporate suitable miniature optical components – filters, mirrors, lenses, etc. with the incidence/collection fibers in the probe head itself instead of keeping them near the source and spectrograph. Many designs discussed below have been studied for probes of this type with filters, lenses, etc., at the probe-tip portion. But the approach has a big disadvantage, namely that such probes with built-in filters can be used with only a single wavelength for incidence and a single spectral range for collection. Raman-scattered/fluorescence radiation is usually coupled into the collection fibers by using a lens to improve efficiency. Such focusing of radiation over a small spot can cause thermal photodegradation of samples like tissue, skin, cellular samples, etc. Another problem in the use of lenses is the energy loss/absorption by reflection/absorption, which can reduce the signal, especially for Raman spectra. This should be minimized by choice of suitable lens material along with anti-reflection coatings for the desired wavelength. Glass fluorescence may also affect the signal since 785 nm incidence wavelength which is commonly employed for biological sample studies, may produce a broad background in the region 1300 and 1800 cm⁻¹. If the low-noise background spectrum is recorded, the fluorescence can be removed by scaled subtraction. Also, the presence of shot noise may decrease the sensitivity of the Raman spectrum. For the proper visualization of the site of interest the camera chip can be added on the tip of probe. In the fiber optic Raman probe, all the optics such as bandpass filter, dichroic mirror, focusing lens, collecting lens, and notch filter can be integrated into the probe. The Raman background from the silica incidence fiber will be scattered from the sample and will also be picked up by the collection fibers, in addition to the sample spectrum. Precautions should be taken to minimize these silica-contributed signals in the recorded spectra. This can be done by the appropriate selection of fiber material and by reducing the incident radiation entering into the collection fibers. Williams et al. [53] investigated the consequence of silica background on the sample spectra and found that the use of low hydroxy (low-OH) silica fibers can reduce the background signal almost by one order of magnitude. Myrick and Angel et al [54]. demonstrated optical facing fiber (OFF) configuration to achieve complete removal of silica Raman background. This configuration involves separate filtering for the incidence fiber and the collection fibers (Figure 2a) [54]. A band-pass filter is used to pass only the laser wavelength from the incidence fiber, and a long pass filter is employed for the collection fibers to pass only the longer wavelengths (stokes-Raman) scattered radiation. This configuration was also adopted by Bello and Vo- Dinh et al [55]. This probe was demonstrated only for liquid samples flowing across the microfluidic channel and will not be useful for solid powders or opaque samples. This problem can be resolved by keeping both the collection and incidence fibers on the same side of the sample at a certain angle to each other, as shown in Figure 2a.

Figure 2. (a) The off configuration and dual filter probe, adapted and redrawn from [59], (b) compact small fiber optic probe to measure near infrared Raman spectra of cervical tissues in vivo adapted from [57], (c) Angled probe for biomedical application, (d) The configuration of fibers in the Raman probe and arrangement of probe tip adapted and redrawn from [56,58], and (e) The fiber-optic probe (made by Visionex) to excite the tissue of walls of blood vessel, with permission of [64], and (f) Optical fiber mini probe for Raman spectroscopy, with permission of [63].

Anita Mahadevan-Jansen et al. [57] developed a compact Raman probe for in vivo measurement of near-infrared (NIR) Raman spectra of cervix for cervical cancer diagnosis (Figure 2b). By using UV-grade silica fibers, suitable filters, and beam-turning optics, they could record good Raman spectra from tissue layers perpendicular to the probe-head to discriminate pre-cancerous and normal tissues. When space is not a constraint (e.g. oral cavity), similar side-viewing can be achieved by bending the probe tip through a small angle (Figure 2 c). Such a probe was used by Unnikrishnan, V. K. et al and Ajeetkumar et al [11,13] for screening of oral malignancy, where fluorescence from 7 different sites in the oral cavity (Buccal mucosa, tongue top, tongue bottom, tongue tip, tongue lateral, palate, and lip underside) were studied. The experimental system mentioned earlier was used for the measurements in the clinic. The fluorescence spectra of different sites in the oral cavity in ‘Normal’ and ‘Malignant’ conditions are very different from each other. It is seen that the different sites of the oral cavity have very different spectral features. With this probe, they could discriminate normal, pre-malignant (leukoplakia, erythroplakia, lichen planus, oral submucosal fibrosis), and malignant tissue sites in vivo, for the different oral cavity sites, with better than 90% sensitivity and specificity.

Shim et al. have designed a Raman fiber optic probe- internally filtered probe (IFP) [58], for biomedical applications, where ‘on-the-tip’ filters were used to improve the collection efficiency of the signal. Few other groups have reported improved collection efficiency and carried out detailed calculations for different probe configurations and numerical modeling of fiber optics probes for specific cases such as Resonance Raman spectroscopy [57,59,60]. The best collection efficiency was achieved when a single fiber is used for dual purposes i.e. both delivery and collection of light, as this configuration provides an entire overlap of delivery and collection light cone [61]. But the single fiber probes have the disadvantage that they are prone to spectral interference from the background signal generated inside the optical fiber.

The IFP, shown in Figure 2d (Gaser light management system, enviva biomedical Raman probes; Visionex Inc., Atlanta, GA), is a 1.5 m long, bifurcated probe with beam steering, filtering, and optical isolation [58]. This probe has a single low-OH central incidence fiber (400 μm core, 0.22NA) and seven collection fibers (300 μm, 0.22NA). One bandpass filter is attached with the incidence fiber to block spontaneous laser emission from a diode laser and Silica Raman lines from probe material, and long-pass filters are used in the collection fibers (Figure 2d) above 25 mm from the probe tip. The distal end cladding walls of the light delivery fiber was coated with aluminum to prevent cross-talk with the collecting fibers by introducing optical isolation with tilted optical fiber axis for better efficiency. In the comparative probe evaluation, Raman spectra recorded from rabbit tissues showed an improved spectral quality for IFP than the unfiltered probe (UFP). This probe showed approximately 3-, 6-, 7-, and 13-fold more signals than the UFP, for muscle, fat, brain, and esophagus tissues, respectively. The rigid probe tip dimensions were 1.4 mm in diameter and 40 mm in length. This type of Raman probe was also used by Robichaux-Viehoever et al. [62] for the in vivo detection of cervical dysplasia. The portable Raman system was used to acquire the Raman spectra of the cervix recorded from 77 subjects (33 dysplasia, 33 normal, and 11 subjects left out with an integration time of 5 seconds.

Conventional diagnosis of lung cancer is performed by ‘bronchoscopy’ inspection of the airway using a fiber optic endoscope. Magee et al [63] carried out ex vivo Raman investigations on normal and malignant lung tissues using a mini fiber-optic Raman probe. This probe was designed in such a way that it can be inserted into the working channel of the bronchoscope and this technique has been suggested as an alternative to the traditional method for lung cancer screening. A photograph of the mini probe (Emvision LLC, Loxahatchee, Florida) for this application is shown in Figure 2f. The probe is composed of one central optical fiber (400 μm diameter) and seven low–OH collection fibers (300 μm diameter) with a total probe diameter of 2.1 mm a bandpass filter is mounted on the probe tip at the light delivery fiber end and a wavelength pass filter is placed at the tip of collection fibers to pass the stokes-Raman signal. An additional long pass filter is also placed after collimation from collection fibers to eliminate Rayleigh scattering signal. The filtered signal is focused onto another optical fiber bundle (each of 100 μm) to be directed to the spectrometer. An improvised version of this Raman probe was demonstrated by Buschman, et al [64] to investigate the side walls of the blood vessel for cardio vascular disease diagnosis. Both in vitro and in vivo intravascular measurements were carried out using the miniaturized optical fiber probe. The fiber-optic probe (Visionex) was used to excite the tissue with ~100 mW (830 nm) radiation. The side-viewing Raman fiber probe used a miniaturized 90° gold plated reflector at the probe tip by means of the epoxy glue. The 90° reflection at the tip of the fiber probe enable the collection of the Raman scattered signal from the walls. This fiber probe was slightly modified in a study by Komachi et al, where they have evaluated coronary atherosclerotic artery using an inner wall of a model blood flow system and was able to detect the phantom target buried inside the wall [65]. It should be noted that the gold-reflector as well as the bend-tip probe discussed earlier, which can look sidewise, have the great advantage that a simple rotation of the probe enables observing multiple adjacent sites in hollow organs like blood vessels, oral cavity, etc. This is of great relevance in many situations, where seemingly nonmalignant tissues adjacent to a malignant site can be examined for field-cancerization effects [66].

Huang et al. [67] developed a fiber probe for a fast dispersive-type near-infrared (NIR) Raman spectroscopy system, to acquire in vivo, noninvasive real-time skin Raman spectra. Studies carried out using a 785 nm diode laser were able to record good quality spectra in real-time. As shown in Figure 3a, this probe has two arms, one for incidence (laser) line delivery toward the skin surface and the other for signal collection. The laser radiation coupled into the fiber (200 μM) is focused onto the skin surface with a spot size of ~3.5 mm with the aid of collimating and focusing optics. The scattered signal from the tissue is collected, collimated with the collimating lens and focused onto the collection fiber bundle consisting of 58 fibers (100 μM) arranged in a circle. At the spectrometer end, this collection bundle fiber is arranged in a parabolic curved shape to eliminate the aberration and improve the signal-to-noise ratio. All the optics are enclosed in a rigid metal body for proper positioning. In vivo Raman spectra of skin, samples were recorded from the palm using this system and good quality spectra could be obtained [68]. Because of the post-fiber optics in this design, it can be used only in applications where the sample surface is at a reasonable distance from the fiber tip. This makes it unsuitable for in vivo spectroscopy of hollow organs. This problem has been solved by using various designs of microendoscopes with fiber-optic probes [31,69,70]. Motz et.al [71] developed a Raman probe for tissue spectroscopy using 830 nm light incidence. As shown in Figure 3b, the central incidence fiber (200 microns, 0.22 NA) in the probe was surrounded with 15 closely packed collection fibers (200 micron, 0.27 NA) for a better collection cone angle. The central light delivery optical fiber was kept within an aluminum jacket to maintain optical isolation and to prevent photon cross-talk with the collection fibers. The light delivery filter and the collection long pass filter was placed in a rod-like arrangement. The overall length of the probe fiber section was 4 m, and a 2 mm, diameter sapphire ball lens was placed at the tip of the probe to focus the light to a tight spot on the tissue. This fiber probe was able to generate good quality signals from in vitro experiments on breast tissue and artery in different pathological conditions.

Figure 3. (a) Fiber Raman probe with two arms: one for incidence and other for collection, adapted from [67,72]with permission from Optica Publishing Group, (b) Design of the Raman probe tip, longitudinal view at the left side and a cross sectional view of fiber filter interface at the right adapted from [56,71]with permission from Optica Publishing Group, (c) The distal end of the MicroRaman probe, adapted from [56,65]with permission from Optica Publishing Group, (d) The design of the MRP, where a lens with a hole size 0.58 mm at the center for incidence source mounted at the tip of probe to improve the efficiency of collection of the probe, adapted from [75]with permission from Optica Publishing Group, (e) A schematic of the fiber Raman system for lung cancer diagnostics, adapted and redrawn from [76,129], (f) Multimodal spectroscopy (MMS) system using the optical fiber spectral probe for simultaneous acquisition of three spectroscopy techniques such as Raman, fluorescence and diffuse reflectance spectroscopy, adapted and redrawn from [79], (g) Cross section and longitudinal view of the probe, with permission of [73,79], and (h) integrated Raman spectroscopy along with trimodal (white-light reflectance, autofluorescence, and narrow-band Raman imaging methods), adapted from [74,81,129]with permission from Optica Publishing Group.

Komachi et al. [65]developed a narrow fiber optic Raman probe of 600 micron in diameter, which can be inserted into coronary arteries. They selected a probe similar to that in [71], but with a reduction in the number and dimensions of the optical fibers. The ball lens was also excluded as shown in Figure 3c. In this work, the central fiber (150 microns) was covered with only eight collection fibers (150 microns). A bandpass filter was mounted at the tip of the delivery fiber and the low pass filter at the tip of eight fibers in a doughnut shaped configuration. The delivery fiber was placed inside a stainless steel sheath to prevent crosstalk with the collection fibers. The experiment performed using 720 nm light showed no damage to coronary artery tissue with laser light transmission of 22 Micro Joule/pulse at 1.8 kHz repetition (40 mW average power). The laser light was focused to the fiber using a 0.25 numerical aperture objective lens that coupled the laser beam into the fiber. The collection fiber was arranged in such a way that all the fibers were vertically arranged fed to the spectrometer end. The image of fiber tips was focused on the slit using a pair of lenses and neutral density (ND) filters. The spectra measured from rabbit blood vessels successfully demonstrated that atherosclerotic lesions in blood vessels, visually indistinctive from normal, can be differentiated from the normal parts. The collection efficiency of this microRaman probe (MRP) was further improved [65] by using a lens with a hole at the center, size 0.58 mm, for incidence source, mounted at the tip of probe (Figure 3d) [75]. The experiments with 785 nm incidence were initially performed on CaCO₃ powder to compare the sensitivity of three different probes, one flat-tipped probe and a lensed probe with two focal lengths (1 mm and 0.58 mm). No significant changes were observed while comparing the lensed probes with different focal lengths, but improvement was evident for the lensed probe with respect to flat tip. The same research team [61] used a single hollow fiber waveguide coated with silver thin film for a dual directional purpose (light incidence and collection of Raman scattered light). The hollow fiber waveguide does not generate Raman scattering and fluorescence signal noise, by itself, during light transmission. This also avoids the need of a complex filtering system at the fiber end as well as spectrometer end, thus improving signal-to-noise ratio (SNR) concerning conventional probe .But the light transmission obtained was only 80% which is not suitable for Raman spectroscopy application where the Raman signal itself is very low, even though signal-to-noise ratio is good, as complete overlap of light cone was possible.

Fiber probe-based Raman system to measure in vivo real-time spectra in lungs was reported by Short et al. [76]. (Figure 3e). Low OH fibers were used to construct the probe. The central incidence fiber (200 µm) was surrounded by a total of 27 collection fibers (100 µm each). The light delivery fiber was covered with the gold-plated jacket to remove the photon cross-talk with the collection fibers. The size of the catheter was 1.8 mm in diameter and 75 centimeters in length. The collected radiation passes through two long pass filter modules and is transmitted to the spectrometer through a fiber bundle containing 54 100-micron fibers packed in a circular geometry in the filter module. The fibers were then directed to the spectrometer in a parabolic arc shape to reduce the aberration as reported [67]. Here two filters were used, as one of the long pass filter had blue transmission for auto-fluorescence imaging and the second for the Raman study. The spectra in the low-frequency region were dominated by the fluorescence emission of hemoglobin in the tissue excited at 785 nm laser. Even then, the signatures corresponding to triolein (triglyceride) were present in the spectra [77]. The fluorescence contribution was less in the high-frequency range (1500 to 3400 cm⁻¹), and the Raman technique was found suitable for the in vivo characterization of abnormal lung tissues with an integration time of 1–2 seconds.

Recently Losch et al. made a patent review for identifying the potential distal end designs for steering light in different directions in medical devices. The review provides three main strategies for light steering in a specific direction such as (i) modifying the total internal reflection pattern inside the fiber, (ii) modifying the optical fiber core end surface, and (iii) adding an optical element outside the optical fiber-based on reflection or refraction, scattering and diffraction. The diffraction with an optical element outside the fiber was found to be suitable for splitting light into multiple beams. For increasing the angular distribution, scattering with an optical element outside the fiber was proposed. They also concluded that the application of refraction or reflection as an optimum process for steering the light beam in desired direction [78].

5. Probes for multi-modal spectroscopy

Šćepanović et al [79]. have designed and developed a multimodal spectroscopic (MMS) system using a single optical fiber probe for the acquisition of Raman, intrinsic fluorescence (IFS), and diffuse reflectance (DRS) in a clinical setting (Figure 3f). The MMS system used three independent incidence light sources and multiplexed them to excite the tissue and collect the spectroscopic information. Raman signals were acquired using 830 nm laser incidence (100 mW), nitrogen laser (337 nm), and a xenon flash lamp (370–740 nm) were used for IFS and DRS measurements, respectively. The MMS probe developed was similar to the fiber Raman probe explained in [80]. The collection fibers were divided into two sections, one each for Raman and DRS/IFS. Ten out of 15 collection probes were utilized to collect the Raman signals and the rest were used for DRS/IFS purposes. All incidence sources were filtered as per the schematic diagram in Figure 3g using appropriate optics and then coupled to a single optical fiber using the optical switch and subminiature version (SMA) connector. This MMS probe was employed for both ex vivo and in vivo investigations to find vulnerable plaques in arteries and also for breast cancer diagnosis. All three spectroscopic signals can be acquired within few seconds by switching between the three incidence sources by computer control. The signal-to-noise ratios in this system were comparable to those where all three types of spectra were collected independently. Huang et al. [81] designed an integrated Raman spectroscopy along with trimodal spectroscopy (white-light reflectance (WLR), autofluorescence imaging (AFI), and narrow-band imaging (NBI)), for real-time in vivo spectroscopy measurements during endoscopy. The Raman spectroscopic system was composed of 785 nm diode laser (300 mW), and an imaging transmission spectrograph. The Raman endoscopic probe with 1.8 mm diameter and 2.5 meters long consisted of one central incidence fiber surrounded by 32 ultralow-OH fused silica optical fibers (200 microns, NA- 0.22) as given in Figure 3h. The band pass filter for the laser incidence source and the long pass filter for the Raman scattered light collection were placed outside the probe before the spectrometer using a proper filtering module. The wide-field endoscopic images such as WLR/AFI/NBI and in vivo Raman spectra of the tissue could be seen in real-time on a computer monitor. Raman spectra of buccal mucosa from healthy subjects were recorded using this setup with an acquisition time of 0.1–1 sec [67,82–84]. Multimodal spectroscopy-based systems are always promising in terms of sensitivity and specificity but these techniques more expensive.

Investigations have also been done with probes, for which, though the optics were not incorporated on the probe itself, the optics and fiber probe have been combined into a single unit, separated from the laser and spectroscopy system. Mo et al. [85] conducted in vivo studies for the detection of cervical malignancy by exploring the Raman features in the high wavenumber region (2800–3700 cm⁻¹). The handheld fiber probe developed for this study (Figure 4a) was made up of two optical arms, one for delivering the 785 nm laser beam (100 mW) to tissue and the other for Raman signal collection. Incidence light is transmitted into the light delivery arm of the Raman probe via a single 200 μm core fiber (NA 0.22), which is further passed through the filtering module. A NIR-coated sapphire ball lens (5 mm diameter) was fixed on the tip of the Raman probe. The backscattered light from the cervical tissue sample up to a depth 700 μm was collected by the ball lens efficiently with a collection cone angles of 55° [85–87] and then reflected by a dichroic mirror (Figure 4a). The collected photons were passed through an edge filter/ long pass filter (cutoff wavelength at 800 nm to block the Rayleigh scattered light) and finally focused into fiber bundle, which carries the light to a spectrometer using a lens. The fiber bundle consisted of 28 optical fibers (50 micron core, NA 0.22) connected to the spectrometer through a line fiber bundle adapter for better signal to- noise ratio using vertical binning to fit the entire CCD [67]. All the above-mentioned optics and connecting fibers of Raman probe were sealed inside a stainless steel sleeve with an outer diameter of 8 mm, and polytetrafluoroethylene (PTFE) gaskets were used for leak proofing in view of in vivo measurements. In vivo Raman spectra in the high wavelength region gave a diagnostic sensitivity of 93.5% and specificity of 97.8% for dysplastic tissue type. Noticeable difference in intensities of Raman bands at 2850 cm⁻¹ and 2885 cm⁻¹ (CH₂ stretching of lipids), 2940 cm⁻¹ (CH₃ stretching of proteins), and the broad band of water (peak at 3400 cm⁻¹in range of 3100–3700 cm⁻¹) were observed in normal and dysplastic cervical tissue. For evaluation of the carcinogenic process in esophageal cancer, Day et al. [88] improvised the Raman probe described above [85] using different optics. The optical design, development and fabrication of this probe used the anisotropic wet etching process technique which is commonly employed or making silicon motherboards and jigs in industries. They considered the probe size one of the most important factors, such that it can pass through the upper gastrointestinal tract and should match with the medical endoscopy instrument (2.8 mm bore size). For this, the rigid, metal part of probe end should not be more than 2 cm in length and the material must be compatible with human anatomy and biochemistry. An optical fiber of 62.5 μm core was used for laser light delivery and a low OH fiber with 105 μm core diameter for collection purposes. The reflecting mirror and long pass interference filter were placed parallel to each other at an angle of 10° to reflect the laser wavelength and to pass the higher wavelengths. This configuration was achieved by incorporating both components into one monolithic unit cut from a conventional block of a 3 mm thick filter [88]. The long pass interference filter is a hard oxide short-pass interference filter to block the fiber fluorescence due to laser incidence which is diced into a 1 mm cube. Another long pass interference filter was used to pass the stokes Raman scattering light, and was made by dicing conventional filter into 1 mm size cubes. Two Gradient Index- GRIN- collimating lenses were used in the incidence and the collection path of the probe. The probe was used to record the Raman spectra from 58 esophageal biopsy samples collected during the endoscopic investigation. They were categorized efficiently into three sub groups such as normal, premalignant and malignant to demonstrate the capability of the fiber Raman probe as an optical biopsy tool. Researchers have tried volume type Raman endoscopic probes to facilitate rapid spectral acquisition of the tissue for cancer detection in different internal organs [81,89].

Figure 4. (a) The optical layout of the hand-held type Raman probe, adapted and redrawn from [73,85],, (b) Schematic of a beveled fiber optic confocal Raman endoscopic probe, adapted and redrawn from [90], (c) Schematic of fiber probe utilized for simultaneous FP/HW region where ball lens coupled with fiber Raman probe spectroscopy, adapted and redrawn from [91], (d) Spectropen the hand held fiber Raman probe, with permission of [95], (e)Multiple fiber probe with GRIN lens and needle bore, adapted and redrawn from [27,96], and (f) Scanning electron micrograph of the Kagome-lattice HC-PCF and experimental setup, adapted from [98]with permission from Optica Publishing Group.

Wang et al., reported the development of a beveled configuration optical fiber-based confocal Raman probe attached with a ball lens (Figure 4b) for improving the in vivo epithelial tissue measurements during endoscopy with respect to the volume type fiber optic endoscopic probes. The confocal Raman probe consists of one central flat tip fiber (200 μm, NA 0.22) for delivering light which is surrounded by beveled face optical fibers (200 μm, NA 0.22) for Raman signal collection. The dependence of parameters such as bevel-angle (β), and gap(d) between fiber tip and ball lens of the confocal probe configuration on collection efficiency were calculated using Monte Carlo (MC) simulation algorithm [90]. Simultaneous increase of fiber-bevel angle from 0° to 25° and the gap from 0 to 600 μm of the confocal fiber Raman probe resulted in a significant decrease in the collection of Raman scattered photons. When the angle was increased from 0° to 20 °, the Raman scattered intensity ratio of the epithelium to stromal tissue has been enhanced by 2.5 times. A maximum epithelium to stroma intensity ratio ~6 was achieved while selecting a beveled angle of 20° at a minimum gap (zero) of fiber and ball lens in the confocal Raman probe design. This was able to collect up to 85% of the total Raman signal generated from the epithelial layer tissue. But the volume type probe could only collect~23% of the total Raman signal originating from the epithelial layer, and the remaining 77% Raman signal was from the deeper stromal layer tissue, suggesting that the volume probe is able to detect deeper tissue content. Duraipandian et al [91] developed for the first time a NIR coated ball lens (diameter 5 mm, n = 1.77) mounted confocal Raman spectroscopy probe (8 mm long) to monitor the variations in the fingerprint (FP) and high-wavenumber (HW) region for in vivo detection of cervical precancer and dysplasia. The system shown in Figure 4c used a 785 nm diode laser (100 mW), spectrometer (QE65000, Ocean Optics) and NIR enhanced back thinned charge-coupled device (CCD) detector to effectively collect the Raman scattered signal from the epithelial tissue. The central fiber (200 μM, 0.22 NA) tip itself was coated with a bandpass filter (central wavelength at 785 nm), and the collection optical fiber (200 μm core, NA = 0.22) was coated with a long wavelength pass filter (cutoff ~800 nm). The coupling sapphire ball lens along with the double fiber Raman probe offered confocality for collection of the Raman signal from epithelial tissue to around~160 μm depth in the human cervix as per the Monte- Carlo simulations [85,90]. The partial least squares-discriminant analysis (PLS-DA) carried out on the Raman spectral data was able to find diagnostically significant Raman spectral components In particular, an increase in nuclear content (1095 cm⁻¹) and a decrease in glycogen and proteins content (854 and 1654 cm⁻¹ respectively) were evident during the dysplastic progression in the cervix.

Wang et al. [92] observed the promising potential for beveled type probes for in vivo detection of gastric dysplasia while comparing the diagnostic performance of beveled and volume Raman endoscopic probes. Mo et al. [90] demonstrated a ball lens mounted Raman probe design configuration for improving depth-selected Raman spectrum measurements of epithelial tissue studies, as shown in Figure 4a. The scattered light collection efficiency can be improved by properly choosing the refractive index and the size of the ball lens for the Raman probe design, as obtained from the Monte Carlo simulations [90].The incidence light was delivered through a single optical fiber (200 μm, NA = 0.22) and bandpass filtering unit. At the tip, a near-infrared (NIR) convex lens was used for collimating incidence light [93]. The overall illuminated light exhibits higher intensity on tissue by using a higher refractive index ball lens, thus improving the Raman scattered light collected from the epithelial layer of tissues. A fivefold increase in Raman scattered intensity was obtained while replacing the lower refractive index lens (1.46) with a higher refractive index lens (1.83). Using the fiber probe mentioned in the above studies [90], Duraipandian et al [94]. measured the in vivo Raman spectra from 57 different cervical tissue sites (35 normal and 22 precancer sites). Total 105 spectra were recorded, from which 65 spectra were from normal sites, and the remaining 40 spectra were from cervical precancerous lesions sites (7 low-grade cervical intraepithelial neoplasia and 33 high grade cervical intraepithelial neoplasia). In this case also, the spectra displayed an increase in nucleic acid contents (979–999 cm⁻¹) and a decrease in glycogen (925–935 cm⁻¹) in precancerous tissues. Mohs et al [95] reported the development of a handheld probe, ‘SpectroPen’ (Figure 4d) for the identification of malignant tumors, which may sometimes be missed by the surgeon during routine procedures. This pen-type probe was used during surgery to measure SERS (surface enhanced Raman scattering) and fluorescence spectra using near-infrared contrast agents for detecting malignant tumors. The probe is of cylindrical shape made up of stainless steel (SS) sampling head (1.3 mm diameter, length 10 cm) and integrated with two 5 meter optical fiber cables. The limits of detection with this probe were obtained as 2–5 × 10⁻¹¹ M for ICG (Indocyanine green) and 0.5–1 × 10⁻¹³ M for the SERS. When excited using 785 nm, ICG displayed a fluorescence peak at 816 nm, while fat (lipids) showed a background fluorescence peak at 805 nm accompanied with resolvable Raman signals at 862, 1070, 1297, 1439, 1652 cm⁻¹.

Day et al [96]. initiated preliminary studies toward the development of a fiber optic Raman probe, which can be incorporated into a standard hypodermic needle, as shown in Figure 4e. The aim was to incorporate the distal tip within a 20 gauge needle having an internal diameter of 0.6 mm. This limited the optical components that can be readily included at the tip. The incidence arm consisted of a 62.5 μm core, low OH fiber to deliver light from a stabilized 830 nm semiconductor laser. The output of the fiber was collimated by a gradient-index (GRIN) lens and transmitted through a short wavelength pass filter. A second, identical GRIN lens was used to couple the light into a 15 cm length of metal-coated, low OH, fiber. The fibers used in the collection path were of 105 or 200 μm core diameter along with a long wavelength pass filter. Plesia et al [97]. explored Raman probe designs to study neuromuscular diseases (amyotrophic lateral sclerosis(ALS), and duchenne muscular dystrophy (DMD) in mouse models. Spectra were acquired over 2 years from multiple groups of mice (i.e. both human mutant superoxide dismutase-1SOD1^G93A/mdx and different ages) tested at each slot. Using the SOD1^G93Amodel of ALS and mdx model of DMD, spectra were collected from the living muscles to observe clear differences from the neighboring tissues such as bone and blood. Ghenuche et al. [98] demonstrated the use of a large-pitch Kagome-lattice hollow-core photonic crystal fiber (HC-PCF) probe for Raman applications (Figure 4f). Due to its large transmission bandwidth, it can be utilized for both incidence and collection. The 632.8 nm laser light was coupled with the 1.5 m long Kagome HC-PCF by a 15 cm focal length lens, which matches with the 0.02 numerical aperture of the HC-PCF. The laser beam exiting the fiber was focused into a sample inside a quartz cuvette by two achromatic doublets with 25 mm focal length and the same lenses were used for back-coupling the Raman scattered light into the fiber. Raman scattered signal was differentiated from the background using a dichroic mirror, a long-pass filter and a notch filter. A small pinhole of 20 μm size was fitted to the fiber output to spatially filter the HC-PCF mode carrying the Raman scattered light. In this case, the silica luminescence background is lowered by two orders than the conventional silica fibers removing the requirement of the fiber background subtraction as here the light is mainly transmitted inside the hollow core of the fiber occupied with air. Here the challenge is to use the probe directly on a sample which can be done by fused shutting the opening of fiber.

Pudney et al. [99] developed a Raman probe without optic fibers to avoid the strong background signal from the fused silica, especially in the fingerprint region. This pen-shaped probe contained a small objective, 12 mm in diameter, specially built for external body application, which can be moved back and forth by a motor to vary the laser focus in the tissue. At the probe tip and in front of the objective, a 0.2 mm-thick fused silica optical window of diameter 6.95 mm was placed against the area to be examined. Good quality Raman spectra were conveniently recorded from different body sites (axilla, scalp and mouth), which were otherwise difficult without significant data loss. The Raman spectra of both scalp and axilla were significantly different from the normal volar forearm skin. O’Brien et al. [100] utilized Raman spectroscopy to evaluate the biochemical changes occurring in the human cervix over the period of pregnancy. The Raman probe (EmVision) shown in Figure 5a (left one) includes inline filtering, with seven low hydroxyl (OH) (0.22 NA,300 micron) collection fibers surrounding one low OH (0.22 NA. 400 micron) fiber which has a band-pass filter positioned in front of it. A donut-shaped long-pass filter was positioned in front of these seven fibers to block the laser light and pass the Raman scattered light from the sample. All these components are enclosed in a stainless steel needle of diameter 2.1 mm. This probe called as the non-superficial Raman probe [101]. They showed that the normalized intensity of amide I (1657 cm⁻¹) reduced noticeably over the course of pregnancy, which was more prominent in the last month of pregnancy. It is well known that the cervical extracellular matrix degrades and loses organization before delivery. The decrease in Amide 1 band can be attributed to this process since amide I bonds are prevalent in extracellular matrix proteins. Using the same external dimensions of the above-mentioned non-superficial probe, another probe was designed with change in the optical components called as superficial Raman probe by (EmVision, USA) shown in Figure 5a (right) [101]. The size of the central incidence fiber was reduced to 200-micron core with 0.22 NA from of 400 micron. For the proper overlapping of collection and incidence cone a unique two-component converging lens was mounted at the tip of the probe. The converging lens was made up of a 1 mm thick plain window of fused silica and a plano convex sapphire lens of diameter 2 mm. The high refractive index of the sapphire lens (~1.7) enables the bending of the light sharply, which creates proper overlapping for maximum signal collection.

Figure 5. (a) The commercial Raman probe (EmVision) includes inline filtering, adapted from [101] with permission from Optica Publishing Group, (b) Visually guided optical Raman probe, with permission of [102], (c) The fiber Raman probe with a small optical window fitted at the sheath cap to maintain constant measurement distance between Raman probe end and tissue surface, with permission of [103], (d) The biopsy needle probe for brain tumor, with permission of [105], (e) The fiber optic probe where laser light was coupled into the incidence fiber which is surrounded by 24 collection fibers with integrated filters at the distal end, adapted from [112] with permission of IOS press, (f) Design of distal end of the fiber probe to be made using ULAE technique, (g) The lenses after laser cutting and polishing of L2 module with permission of [113], and (h) Schematic setup of SERS fiber probe and the enhanced graphical image of fiber tip, with permission of [114].

O’Brien et al. [102] presented a visually guided optical Raman probe for the assessment of biochemical changes throughout pregnancy is shown in Figure 5b. The device incorporated a Raman spectroscopy probe for biochemical monitoring and a camera for visualizing measurement locations to avoid the locations with cervical mucus and blood. The experimental results acquired from patients with and without a speculum showed similar spectra, indicating the potential of visually guided probe in conserving data quality. The Raman spectroscopy probe was made up of a 300 μm low OH fiber for incidence, surrounded by seven 300 μm fused silica collection fibers (slightly displaced from the illumination fiber), all contained in a 2.1 mm stainless steel tube. A microlens located at the distal end of the probe (3 mm, sapphire plano convex lens) coupled to a MgF₂ window enabled incidence and collection cone overlap deep within the sample. An adjacent stainless steel tube containing a 1 × 1 x 1.7 mm CMOS detector, which has 120 degree field is placed for visual guidance, and a silica window is placed in front of it to remove the autofluorescence and Raman scattering due to camera unit. Wang et al. [103] designed a probe configuration, where a small optical window ~ 1 cm thick was attached to maintain constant distance between Raman probe end and tissue surface (Figure 5c). The 3-meter-long Raman endoscopic probe (2.3 mm diameter) [104] was made up of a single incidence fiber (200 micron) surrounded by 34 collection fibers (100 micron) to collect the Raman signal. To ensure that the sheath caused a minimal amount of fluorescence and Raman interference, Raman probe was enclosed in a stainless steel mono coil to provide sufficient tensile strength and enough flexibility. The protective sheath was used over the Raman endoscopic probe to prevent the loss of signal to due to the aging of the probe. The probe was covered with a medical grade heat shrink tubing made up of Iridium, which protected it from biological contaminants. Transparent polyvinylidene difluoride (PVDF) material with high tensile strength was used to cover the head of Raman probe. The probe was tested using tylenol tablet without much loss of signal and it was found suitable for in vivo spectroscopy application.

Desroches et al. [105] used label-free Raman spectroscopy, based on high wavenumber Raman spectroscopy, to analyze the molecular nature of the tissue for cancer targeting and detection before removing tissue for biopsy during the surgery. This device was engineered into a commercially available biopsy system allowing tumor analysis prior to tissue harvesting without disrupting workflow. The optical needle biopsy with an enlarged view of the tip showing the biopsy window and the beveled optical fibers used for incidence and collection fibers for the detection of biomolecular markers is shown in Figure 5d. In use, the needle is inserted into the examining area, Raman spectrum is, recorded, and the biopsy window is rotated to cut/remove and collect the tissue sample. The biopsy needle probe shown in Figure 5d comprises 12 low-hydroxyl fibers (105 μm core diameter, 0.22 NA) for incidence and collection of Raman signal. The human brain study was conducted using this probe to find the difference in the Raman spectra in high wavenumber region in 19 patients with grade 2–4 glioma during the surgery. The protein peak at 2930 cm⁻¹ was found to be more intense for the dense cancer tissue compared to normal brain tissue, which also increased protein/lipid ratio i.e. the 2930/2845 cm⁻¹ range, for the dense cancer tissue. In vivo Raman spectroscopy suffers from a few but yet significant drawbacks [86,106]. Tissue often exhibits fluorescence background which is superimposed on the Raman spectra. For highly fluorescent tissues such as liver, kidney and lung, the background level is still not acceptable for subsequent spectral analysis and needs to be corrected [107]. The scattered light and collection geometry of most Raman fiber probes is divergent and non-confocal [108]. The ambient light present during surgical procedures is also a major hindrance for effective spectral analysis [109]. In the wavelength-modulated Raman spectroscopy, ambient light and system transmission function do not significantly change, whereas the Raman signals do vary upon multiple wavelength incidence with small wavelength shifts [110,111]. Wavelength modulated Raman spectroscopy used in the study by S. Dochow et al. was able to overcome the key problems associated with conventional Raman spectroscopy [112]. In this study, the maximum output power of 785 nm light was reduced to 150 mW from 1 W at the 10X microscope objective output with an effective NA of 0.25. One incidence fiber surrounded by 24 collection fibers with integrated filters at the distal end used in the study is shown in Figure 5e. The modulated method allowed recording Raman spectra even in highly fluorescent liver tissues, and further simplification in data processing may be used to improve classification for in vivo diagnosis.

Ross et. Al. designed and developed a miniature Raman probe with the aid of ultrafast laser assisted etching (ULAE) technique [113]. Six collection fibers were used alongside a single incidence fiber, as shown in schematic Figure 5f. The optical fibers were paired with a flower petal shaped lens array shown in Figure 5g inside the L 2 module. One central plano convex aspheric lens for incidence and six plano convex aspheric lenses of 0.8 NA were embedded inside the L2 module. The central lens only occupied 4.2% of the scattered light profile leaving ~ 96% to be collected by six surrounding lenses. Around 85.1% of the scattered light was finally collected after the loss due to Fresnel reflection at the optical surfaces was determined by ray-trace simulations. Ultrashort laser pulses were used to modify transparent material’s chemical and physical properties locally within a small well-confined focal volume using three-dimensional micromanufacturing by (ULAE). Using the ULAE, the fused silica material was modified using controlled laser irradiation and subsequent etching to fabricate the set of incidence and collection lenses. The six collection fibers were arranged in linear array and coupled to the spectrometer by aligning the image of fibers at the slit of the spectrometer. The long pass filter was placed at the return path of light to the spectrometer. The efficiency of Raman probe was initially verified using standard samples like isopropanol, toluene and silicon and later on ex-vivo mouse intestine tissue. The study using the probe didn’t find any significant difference between benign and malignant intestine tissue for the mouse model.

A surface enhanced Raman spectroscopy (SERS) fiber optic Raman plasmonic probe using repeated de-wetting was developed by Kwak et al [114]. The nanogap-rich gold nano islands were formed on a single multimode silica optical fiber with repeated de-wetting, after deposition of gold through evaporation in pressurized chamber. The same fiber was used for incidence and collection purpose. A schematic of the setup is shown in Figure 5h. The SERS measurement of the BT (Bovine turbinate) molecule dissolved in methanol was done by immersing the fiber probe in the solvent. The intensity of peak at 1069 cm⁻¹ was used to calculate the enhancement factor (EF) for the SERS and found to be 7.8*10⁶. Repeated de-wetting showed 10.4-fold more enhancement than single de-wetting and this probe has demonstrated high potential for in vivo application.

An overview of the major types of the optical fiber probes have been discussed along with their configurations and some important results. The types of probes, the basic design aspects and their applications are summarized in Table 1.

Table 1. Different configuration of Raman probe designs for various applications.

Fiber optic Raman probe type	The basic probe design and configuration with fibers	Biomedical application/ sample type	Reference
High-axial resolution hollow waveguide probe	Hollow optical fiber of core 320 µm, with ball lens. Probe head diameter 640 µm	Stomach tissue of mice in vitro	[115]
Optical fiber probe for less Background	A silica cladding silica-core fiber (with 100 micron core NA = 0.22) for incidence and single silica-cladding-silica-core optical fiber (600 micron, NA = 0.22) used as collection fiber	Not available/ Benzoic acid, p-nitro phenol and methyl red	[116]
Visually guided Raman probe	Seven low-OH silica (300 µm) fiber surrounded by one low-OH silica (300 µm) fiber with visual camera, outer diameter is 7.1 mm stainless steel	In vivo cervix tissue	[102]
Hollow core optical waveguide	No conventional fiber was used, single hollow waveguide used for minimizing background	Not available/ CaCO3	[61,65]
Envivia by Visionex (commercial probe)	Inline filters in tip to reduce silica signal interference, 6 fibers surrounded by one single incidence fiber	Measurement of Raman spectra of different ex vivo tissues of rabbit such as muscle, brain, skin etc. In vivo skin of human body also recorded	[58,117]
fiber Raman probe for human cervix measuremet	Fiber optic probe bundle with filters (Probe size 12 mm in diameter)	In vivo Raman spectra measurement of normal and precancerous cervical tissue	[57]
Fiber optical Raman probe	2 mm ball lens mounted probe and diameter of probe is 2 mm	Malignant and normal breast tissue	[71]
MicroRaman fiber optic probe	MicroRaman Endoscope- microRaman probe of size 0.6 mm in diameter equipped with 2.5 mm endoscope	In vivo Raman rat esophagus and stomach tissue	[118]
Micro Raman fiber endoscope probe for esophageal track Raman measurement	Confocal probe used various lens for used as objective. The multimode fiber of 62.5 mm core as incidence and low OH, stepped index fiber of 105 μm core diameter (0.22NA)	Ex vivo implementation in Human Barrett’s esophagus for esophageal cancer	[88,119,120]
Commercial EMvision probe	Six collection fiber around one incidence fiber, Filter employed by coating the fiber tip and lenseing for beam steering, filters are arrnged inside the probe itself	Used in vivo during the brain cancers surgery. In vivo atherosclerotic plaque depositions of rabbit. , ex vivo study of tissue for lung cancer	[63,101,121–124]
Spatially offset Raman spectroscopy (SORS)probe with multiple offsets	Single 200 micron core fiber for incidence and collection individually with band pass and long pass filters at fiber tip respectively.	Human soft layer tissue for differntiation of tissue type, and for the ex- vivo detection of tumor of breast	[125,126]
Endoscopic probe with less depth of field	1.8-mm Raman fiber optic endoscopic probe, which can be placed at the biopsy channel of medical endoscopes, 32 Raman collection fibers of 200 micron core surrounded by one central laser delivery fiber of 200 micron core	In vivo Raman spectroscopic measuremet of gastric dysplasia tissue	[81,127]
Multicore fiber with integrated fiber	Potential to be semi-disposable Bragg gratings	Ex-vivo study of brain tissue	[128]

6. Prospects and challenges

At present there are still many challenges in the use of present fiber probe systems, few due to the inherent properties of the different spectroscopic processes and others due to currently available fiber optic systems and optical components. A major drawback/advantage of fiber optic systems is that while they have the limitation of capability of viewing only small areas at a time (limited by the numerical apertures of typical probes, usually about 0.22) this is also an advantage in many cases, especially for tissue spectroscopy, where tissue characteristics can change across nm dimensions. Another disadvantage of fiber probes, due to the small numerical aperture, is the relative inefficiency of the system in Raman spectroscopy, where only a very small fraction of the total Raman signal, which is inherently weak can be collected, even with best matching optics, while laboratory systems can measure much more scattered intensity, by using large N.A optical components. This problem is further accentuated by the need to use NIR wavelengths (785, 830, 1064 nm) to avoid the much stronger tissue fluorescence from use of visible or near-UV radiation for observation of Raman spectra, which again sacrifices Raman intensity due to the dependence of Raman scattering inversely on the 4^th power of exciting radiation wavelength. But this disadvantage also can be reduced by using techniques like time gated spectroscopy, in which since one can use pulsed-laser-time-gated system to separate the instantaneous Raman signal from the delayed fluorescence signals [32,34]. Since almost all biological molecules have electronic absorption in the 200–250 nm region, one can then use the Resonance Raman effect advantageously, in addition to the wavelength advantage from much shorter incidence wavelengths.

Another disadvantage of using optical fibers is that the polarization properties of radiation in both light delivery and collection fibers will be scrambled, thus making them unsuitable to measure the depolarization ratio in Raman spectra (which depends upon the symmetry of the molecule and the normal vibrational modes) for characterization of the band types in different samples of same composition, differing in surroundings, complex formation etc.

In addition to the advantages already mentioned, there are other advantages in using optical fiber probes for spectroscopy studies. One major advantage is that, because the fiber excites/collects signal only from its direct point of incidence/observation, the technique can be efficiently used for point-of-use studies, like Spatially Offset Raman Spectroscopy (SORS), AFM-Raman etc. In short, a specific fiber probe design is often more suitable for a chosen application, and hence different types of probes are strongly recommended for various dedicated applications. For example, different constraints exist when a fiber probe is to be used within a cardiovascular catheter compared to its use in the skin or oral cavity. So, parameters such as the design of the filters, the collection system, and the beam steering properties are important. Most importantly, the shape, size, probe diameter, and flexibility of the fiber probe, all should be considered when one wants to use an Optical Fiber probe. For in vivo applications, the probe materials need to be biocompatible to avoid toxic effects, and suitability for hospital sterilization also should be taken into account for repeated use [129].

7. Conclusion

This manuscript deals with the application of optical fiber probes multiplexed with Raman spectroscopy for in vivo applications. The focus of the study is on the origin and evolution of the optical probes for in vivo application, revealing the progress from single mode/multimode fiber-based system with single spectroscopic technique to multimodal spectroscopy. The efficiency of the probe was found to be increased with the use of multiple fiber system with increased number of collection fibers based on their varying geometries, incidence fiber and cladding thickness. Even though the incorporation of optical elements such as filters in the fiber itself support the attempt to transform the fiber-based spectroscopic technique from bench side to bed side in view of its compactness, but it is restricted due to the use of multiple incidence sources. Optical fiber-basedmultimodal spectroscopic probes have been increased the sensitivity of the technique and can be considered as a potential candidate for future applications. Irrespective of the progress achieved by optical fiber-based spectroscopic techniques for in vivo application, the technology needs to be improved further with more focus on efficient light delivery and collection system in a cost-effective manner.

8. Expert opinion

The high cost of currently available diagnostic methods (CT Scan, MRI, Pathological examinations, etc.) and the non-availability of the large number of qualified medical professionals needed to operate such facilities at a large number of locations have made universal health care, in-accessible, unavailable, and unaffordable to the general population in low/middle (L-M) income countries, where advanced health-care facilities are very limited, being available mostly only in big cities, while a large fraction of the population, almost 70%, reside in small towns and villages. The situation can be improved by providing cost-effective, portable-/semi-portable systems, which can be operated by trained technicians on a POC basis and diagnosis can be arrived at by Artificial Intelligence (AI).Machine Learning (ML) data processing methods, eliminate the need for medical professionals in a primary, real-time screening and detection process.

Spectroscopic methods with optical fiber probes are getting more and more attention at present because of the increased awareness of need for providing universal health care, in a cost-effective, POC basis. Fiber optic systems have made considerable advances recently, because of the highly improved quality of optical fibers used for communication needs. Advances in miniaturization of spectroscopic instrumentation, detectors, and radiation sources like small multi-wavelength lasers and LEDs have now made it possible to use miniature, portable fiber-optic-based spectroscopy instrumentation for screening and diagnosis of a variety of health conditions for universal health care.

In vivo/POC, spectroscopy in an ‘As Is Where Is’ condition, has now reached the stage where various spectroscopic methods can be used for material characterization, in an unambiguous manner, with Fiber-optic systems. At present, Hyperspectral imaging with ‘Chip-on-Tip’ technology on fiber-optic probes, has been shown to be very effective in widely different clinical applications like discrimination of tissue components, (nerves, veins, different tissues, clusters of cancerous cells, perfusion extend etc.), digital pathology, imaging at single-cell or large area level, surgical boundary demarcation etc. so that minimally invasive medical procedures, can be carried out with minimum damage to the organs, enabling faster recoveries [130]. The optical fiber probe-based devices are cheaper, compact and portable, compared to laboratory-based systems. Till recently, for various applications the probe shape, size, material, and probe flexibility had to be customized, because the probe end itself had to contain the dichroic mirror, band pass filter, and low pass/ notch filter etc. making the probe bulky and may be cumbersome. The simplest configuration can be the one where all the filters and the mirrors are placed outside the probe. The simple configuration can significantly reduce the overall cost of the system. The requirement of sterilization after each use can be overcome with the use of disposable cap. At present Hyperspectral imaging systems have already started to be commercially available [130], and fluorescence-based devices are in the advanced stage of technology readiness level (TRL). Raman-based probes are still in their final development stage [131].

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Acknowledgement

Authors are thankful to VGST (GRD No. 459) Govt. of Karnataka, for the sanction of fund to establish the biophotonics laboratory under the scheme, Centre of Excellence in Science, Engineering and Medicine (CESEM), Department of Biotechnology (DBT), Govt. of India for the micro-Raman spectroscopy laboratory facility and DST-FIST, Govt. of India, for other facilities. Authors are thankful to Mr Vittal Shenoy and Mr Udaya for their support. Mr. Ajaya Kumar Barik and Mr. Sanoop Pavithran M are thankful to Dr. TMA Pai Ph.D. Scholarships of Manipal Academy of Higher Education, Manipal.

Reviewers Disclosure

Peer reviewers on this manuscript have no relevant financial relationships or otherwise to disclose.

Additional information

Funding

This paper was not funded.