Tags

Type your tag names separated by a space and hit enter

OctNET: A Lightweight CNN for Retinal Disease Classification from Optical Coherence Tomography Images.
Comput Methods Programs Biomed. 2021 Mar; 200:105877.CM

Abstract

BACKGROUND AND OBJECTIVE

Retinal diseases are becoming a major health problem in recent years. Their early detection and ensuing treatment are essential to prevent visual damage, as the number of people affected by diabetes is expected to grow exponentially. Retinal diseases progress slowly, without any discernible symptoms. Optical Coherence Tomography (OCT) is a diagnostic tool capable of analyzing and identifying the quantitative discrimination in the disease affected retinal layers with high resolution. This paper proposes a deep neural network-based classifier for the computer-aided classification of Diabetic Macular Edema (DME), drusen, Choroidal NeoVascularization (CNV) from normal OCT images of the retina.

METHODS

In the proposed method, we demonstrate the feasibility of classifying and detecting severe retinal pathologies from OCT images using a deep convolutional neural network having six convolutional blocks. The classification results are explained using a gradient-based class activation mapping algorithm.

RESULTS

Training and validation of the model are performed on a public dataset of 83,484 images with expert-level disease grading of CNV, DME, and drusen, in addition to normal retinal image. We achieved a precision of 99.69%, recall of 99.69%, and accuracy of 99.69% with only three misclassifications out of 968 test cases.

CONCLUSION

In the proposed work, downsampling and weight sharing were introduced to improve the training efficiency and were found to reduce the trainable parameters significantly. The class activation mapping was also performed, and the output image was similar to the retina's actual color OCT images. The proposed network used only 6.9% of learnable parameters compared to the existing ResNet-50 model and yet outperformed it in classification. The proposed work can be potentially employed in real-time applications due to reduced complexity and fewer learnable parameters over other models.

Authors+Show Affiliations

Department of Electronics and Communication Engineering, National Institute of Technology Tiruchirappalli, Tamilnadu-620015, India. Electronic address: sunijaprabhakaran@gmail.com.Department of Electronics and Communication Engineering, National Institute of Technology Tiruchirappalli, Tamilnadu-620015, India. Electronic address: research@saikatkar.me.Department of Electronics and Communication Engineering, National Institute of Technology Tiruchirappalli, Tamilnadu-620015, India. Electronic address: gsgayathriunnithan@gmail.com.Department of Electronics and Communication Engineering, National Institute of Technology Tiruchirappalli, Tamilnadu-620015, India. Electronic address: varun@nitt.edu.Department of Electronics and Communication Engineering, National Institute of Technology Tiruchirappalli, Tamilnadu-620015, India. Electronic address: palan@nitt.edu.

Pub Type(s)

Journal Article

Language

eng

PubMed ID

33339630

Citation

A P, Sunija, et al. "OctNET: a Lightweight CNN for Retinal Disease Classification From Optical Coherence Tomography Images." Computer Methods and Programs in Biomedicine, vol. 200, 2021, p. 105877.
A P S, Kar S, S G, et al. OctNET: A Lightweight CNN for Retinal Disease Classification from Optical Coherence Tomography Images. Comput Methods Programs Biomed. 2021;200:105877.
A P, S., Kar, S., S, G., Gopi, V. P., & Palanisamy, P. (2021). OctNET: A Lightweight CNN for Retinal Disease Classification from Optical Coherence Tomography Images. Computer Methods and Programs in Biomedicine, 200, 105877. https://doi.org/10.1016/j.cmpb.2020.105877
A P S, et al. OctNET: a Lightweight CNN for Retinal Disease Classification From Optical Coherence Tomography Images. Comput Methods Programs Biomed. 2021;200:105877. PubMed PMID: 33339630.
* Article titles in AMA citation format should be in sentence-case
TY - JOUR T1 - OctNET: A Lightweight CNN for Retinal Disease Classification from Optical Coherence Tomography Images. AU - A P,Sunija, AU - Kar,Saikat, AU - S,Gayathri, AU - Gopi,Varun P, AU - Palanisamy,P, Y1 - 2020/11/28/ PY - 2019/12/10/received PY - 2020/11/22/accepted PY - 2020/12/20/pubmed PY - 2021/5/15/medline PY - 2020/12/19/entrez KW - Artificial intelligence KW - Class activation mapping KW - Computer-aided detection and diagnosis KW - Eye KW - Machine learning KW - Optical coherence tomography SP - 105877 EP - 105877 JF - Computer methods and programs in biomedicine JO - Comput Methods Programs Biomed VL - 200 N2 - BACKGROUND AND OBJECTIVE: Retinal diseases are becoming a major health problem in recent years. Their early detection and ensuing treatment are essential to prevent visual damage, as the number of people affected by diabetes is expected to grow exponentially. Retinal diseases progress slowly, without any discernible symptoms. Optical Coherence Tomography (OCT) is a diagnostic tool capable of analyzing and identifying the quantitative discrimination in the disease affected retinal layers with high resolution. This paper proposes a deep neural network-based classifier for the computer-aided classification of Diabetic Macular Edema (DME), drusen, Choroidal NeoVascularization (CNV) from normal OCT images of the retina. METHODS: In the proposed method, we demonstrate the feasibility of classifying and detecting severe retinal pathologies from OCT images using a deep convolutional neural network having six convolutional blocks. The classification results are explained using a gradient-based class activation mapping algorithm. RESULTS: Training and validation of the model are performed on a public dataset of 83,484 images with expert-level disease grading of CNV, DME, and drusen, in addition to normal retinal image. We achieved a precision of 99.69%, recall of 99.69%, and accuracy of 99.69% with only three misclassifications out of 968 test cases. CONCLUSION: In the proposed work, downsampling and weight sharing were introduced to improve the training efficiency and were found to reduce the trainable parameters significantly. The class activation mapping was also performed, and the output image was similar to the retina's actual color OCT images. The proposed network used only 6.9% of learnable parameters compared to the existing ResNet-50 model and yet outperformed it in classification. The proposed work can be potentially employed in real-time applications due to reduced complexity and fewer learnable parameters over other models. SN - 1872-7565 UR - https://www.unboundmedicine.com/medline/citation/33339630/OctNET:_A_Lightweight_CNN_for_Retinal_Disease_Classification_from_Optical_Coherence_Tomography_Images_ L2 - https://linkinghub.elsevier.com/retrieve/pii/S0169-2607(20)31710-7 DB - PRIME DP - Unbound Medicine ER -