Deep CNN based microaneurysm-haemorrhage classification in retinal images considering local neighbourhoods

ABSTRACT

Retinal image characteristics can be utilised for early diagnosis of Diabetic Retinopathy (DR). The earliest symptom of DR is the presence of microaneurysm and haemorrhage in retinal fundus images. A computerised classification system can increase the effectiveness of the large volume screening process of retinal images. In this paper, a deep convolutional neural network-based pixel classification approach has been presented where the models, trained on online public datasets, are used for symptom level classification of the fundus images, collected from patient data of a local state hospital. The method achieves average values of sensitivity of 0.4556, specificity of 0.8395 and accuracy of 0.8341 on the local dataset. The CNN did not require exhaustive training with a large number of images as symptom level training was performed on annotated overlapping patches. The average classification time is 0.7275 sec/image. The results in terms of classification metrics and in terms of execution time requirement are very encouraging when compared with different recently developed classification methods and as per the simplicity of the method is concerned.

KEYWORDS:

• Micro aneurysm (MA)
• haemorrhage (HM)
• convolutional neural network (CNN)
• AUC
• sensitivity
• computational time

Declarations

Disclosure statement

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

Availability of Data and Material

The manuscript has data associated with online data repositories.

Ethics approval

The manuscript or any significant part of it is not under consideration for publication elsewhere, nor it has appeared elsewhere as a prior or duplicate publication of the same.

Additional information

Funding

The authors have no funding to report.

Notes on contributors

Mahua Nandy Pal

Mahua Nandy Pal is an M.Tech in Computer Science and Engineering and works as Assistant Professor in Computer Science and Engineering Department of MCKV Institute of Engineering, Howrah, India. Her areas of interest are Deep Learning, Image Processing and Retrieval, Pattern Recognition, Quantum Image Processing and Quantum Learning. She has a good number of publications in the relevant fields.

Ankit Sarkar

Ankit Sarkar is a B.Tech in Computer Science and Engineering and works as System Engineer in Tata Consultancy Services, Hyderabad, India. His areas of research interest are Deep Learning, Pattern Recognition and Analysis. He has some publications and initiated several communications in the relevant fields.

Anindya Gupta

Anindya Gupta is an MS in Ophthalmology and works as Associate Professor, Regional Institute Of Ophthalmology, Medical College, Kolkata, India. His areas of research interest are community ophthalmology, accommodative anomaly and ametropia. He has publications in the fields.

Minakshi Banerjee

Minakshi Banerjee is a Doctorate in Computer Science and Engineering and works as Professor in Computer Science and Engineering Department of RCC Institute of Information Technology, Kolkata, India. Her areas of research interest are Deep Learning, Image Retrieval, Pattern Recognition and Analysis. She has a lot of good publications in the relevant fields.