Robust semi-automatic segmentation method: an expert assistant tool for muscles in CT and MR data

ABSTRACT

Image muscle segmentation is useful to quantitatively assess musculoskeletal diseases by extracting biomarkers such as shape, texture and water diffusivity metrics. Although volumetric manual segmentation is time consuming and a bottleneck in practice, fully automatic approaches are still in progress to reach an acceptable accuracy. In this paper, we provide a robust semi-automated tool to segment two musculoskeletal systems, i.e. thigh and shoulder in MRI and CT modalities, respectively. The tool only needs a few manually labelled cross-sections to build a directed graph-structure of corresponding points between the successive spaced slices. The boundaries of each muscle are obtained by performing a spline interpolation based on the directed graph-structure. Each muscle label and its corresponding 3D mesh are deduced using post-processing techniques. We evaluated the tool on 26 MRI thighs and 16 CT shoulders. Three metrics along with inter-muscle overlapping were employed to evaluate the tool by comparison to an expert manual segmentation and a publicly available tools (ITK-SNAP, 3D Slicer). The results showed a mean Dice $0.988\pm 0.003$, and Hausdorff Distance $4.86\pm 1.67$ mm in comparison to the manual reference for thigh muscle segmentation, and a mean Dice $0.961\pm 0.005$ and Hausdorff Distance $2.42\pm 0.79$ mm for shoulder muscle segmentation, outperformed the other methods. The tool is proposed as slicer module available at https://github.com/latimagine/SlicerSpline.

KEYWORDS:

• Thigh muscle
• shoulder muscle
• 3D spline interpolation
• directed graph-structure

Acknowledgment

This work was supported by French grant managed by the National Research Agency under the “programme d’investissements d’avenir” bearing the reference ANR-17- RHUS-0005 (Followknee), and the financial support of the Brittany Region. The tools was registered by “Agence pour la protection des programmes” with number IDDN.FR.001.420002.000.S.P.2021.000.21000.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21681163.2023.2301403

Additional information

Funding

The work was supported by the Agence Nationale de la Recherche [ANR-17- RHUS-000].

Notes on contributors

Mehran Azimbagirad

Mehran Azimbagirad (PhD) is a research fellow in medical image processing at the University College London and former postdoc in the University of Western Brittany, France. He received his BSc and MSc degrees in applied mathematics and numerical analysis from Razi University and his PhD in applied physics in medicine and biology from University of Sao Paulo, Brazil, in 2019. Medical image processing, mathematical modelling and machine learning are his main research fields.

Guillaume Dardenne

Guillaume Dardenne (PhD) is a Research Scientist at INSERM, the French National Institute of Health and Medical Research, and is in charge of the Computer-Assisted Orthopedic Surgery (CAOS) research team within the Laboratory of Medical Information Processing (LaTIM, France). His research mainly concerns the development of software-based solutions to better personalize patient surgical management by considering all the different phases: (1) during the planning through the automatic analysis of multimodal information for the accurate patient modelling, (2) during surgery by providing non-intrusive systems to better assist the surgical team, and (3) during the follow-up with the design of connected implants and infrastructure to further monitor patient recovery. He is currently co-scientific leader of the FollowKnee project (ANR - RHU, €24M) for the knee surgery, is in charge of the PLaTIMed platform (www.platimed.fr) for testing innovative interventional devices and is involved in scientific societies in the field, “CAOS International” and “CAOS France”, whose aim is to promote the use of new technologies in orthopedics.

Douraied Ben Salem

Douraied Ben Salem, M.D., Ph.D., Habil., is Professor of Radiology at the University of Brest (France) and the Editor-in-Chief of the Journal of Neuroradiology. He received his medical and doctoral degree from the University of Burgundy (Dijon, France). He is the head of the first program in the world of Gadolinium recycling called MEGADORE (MEdical GADOIinium REcycling) and the Vice chair of the ESNR (European Society of NeuroRadiology) Green Committee.

Jean-David Werthel

Jean-David Werthel is associate professor at Hôpital Ambroise Paré where he is responsible for the shoulder unit. He completed his training in Paris and spent a year as a research fellow in the biomechanics lab of the Mayo Clinic. He also completed a PhD in computer science at the IMT Atlantique in partnership with Imascap. His interests are in complex reconstructive surgery of the shoulder and computer-assisted surgery as well as in basic science research related to the biomechanics of the shoulder and of shoulder implants. He is involved in the design of implants for shoulder replacements and has presented his research on numerous occasions at national and international meetings. He is an active member of the Chaine de l’Espoir, a French nonprofit organization working with disadvantaged children, with which he performs regular surgical missions for obstetric brachial plexus injuries and hand congenital deformities. Jean-David has published over one hundred and twenty scientific articles and is a reviewer for shoulder articles and corresponding member in several journals.

François Boux de Casson

François Boux de Casson is the scientific evidence program manager at BlueOrtho, a subsidiary of Exactech. He received his BSc degree in mechanics, his MSc degree in computer science and his PhD in computer science from University of Grenoble. He worked in the medical device field for 20+ years, developing computer assisted orthopaedic solutions.

Eric Stindel

Eric Stindel received the Ph.D. degree in biomedical engineering in 2003 from the Joseph Fourier University, Grenoble, France.,He is currently an Orthopaedic Surgeon at the University Hospital of Brest, Brest, France. He joined the Laboratory of Medical Information Processing, French National Institute of Health and Medical Research, Brest, France, in 1996, where he is a Full Professor. He leads the group which focuses on physiological information for the musculoskeletal system. His research interests include medical image processing, motion analysis, and computer-assisted surgery.,He is a member of the French Biomedical Engineering Society and a Founder of the French Association for Computer-Assisted Orthopaedic Surgery.

Charles Garraud

Charles Garraud is a R&D engineer in medical image processing at the Laboratory of Medical Information Processing in Brest. He received his MSc degree in telecommunication and signal processing from University of Bristol and worked for many years in the medical device industry.

Olivier Rémy-Néris

Olivier Rémy-Néris is full professor at the faculty of medicine of the University Brest, France. He is the head of the rehabilitation department at the university hospital in Brest. He received his PhD in biomechanics from the university of Paris Saclay, in 2001. Biomechanical biomarkers of the rehabilitation process in several pathologies are his main research field.

Valérie Burdin

Valérie Burdin a full professor at IMT Atlantique, conducts her research activities in the INSERM UMR 1101 LaTIM, in collaboration with both the orthopedic surgery and physical medicine and rehabilitation departments of the Brest University Hospital. Her expertise concerns the 3D morpho-functional modelling of musculoskeletal systems, with a particular focus on multi-object statistical modelling encapsulating shape, movement, and intensity. Valérie Burdin (H-index: 15) is author or co-author of over 140 peer-reviewed publications