Deep learning architecture "LightOCT" for diagnostic decision support using optical coherence tomography images of biological samples.

Ankit Butola,Anurag Srivastava,Dilip K Prasad,Vishesh Dubey,Dalip Singh Mehta,Azeem Ahmad,Darakhshan Qaiser,Paramasivam Senthilkumaran,Balpreet Singh Ahluwalia

doi:10.1364/boe.395487

Abstract

Optical coherence tomography (OCT) is being increasingly adopted as a label-free and non-invasive technique for biomedical applications such as cancer and ocular disease diagnosis. Diagnostic information for these tissues is manifest in textural and geometric features of the OCT images, which are used by human expertise to interpret and triage. However, it suffers delays due to the long process of the conventional diagnostic procedure and shortage of human expertise. Here, a custom deep learning architecture, LightOCT, is proposed for the classification of OCT images into diagnostically relevant classes. LightOCT is a convolutional neural network with only two convolutional layers and a fully connected layer, but it is shown to provide excellent training and test results for diverse OCT image datasets. We show that LightOCT provides 98.9% accuracy in classifying 44 normal and 44 malignant (invasive ductal carcinoma) breast tissue volumetric OCT images. Also, >96% accuracy in classifying public datasets of ocular OCT images as normal, age-related macular degeneration and diabetic macular edema. Additionally, we show ∼96% test accuracy for classifying retinal images as belonging to choroidal neovascularization, diabetic macular edema, drusen, and normal samples on a large public dataset of more than 100,000 images. The performance of the architecture is compared with transfer learning based deep neural networks. Through this, we show that LightOCT can provide significant diagnostic support for a variety of OCT images with sufficient training and minimal hyper-parameter tuning. The trained LightOCT networks for the three-classification problem will be released online to support transfer learning on other datasets.

Highlights

Optical coherence tomography (OCT) is emerging as an increasingly popular technique, which is capable of capturing microscopic and real-time imaging of tissues without exogenous contrast agents
We show that LightOCT can provide significantly high accuracy for cancer detection and classification between different types of retinal diseases
We propose LightOCT, a convolutional neural networks (CNN) architecture, can provide an excellent accuracy for different OCT image datasets

Summary

Introduction

Optical coherence tomography (OCT) is emerging as an increasingly popular technique, which is capable of capturing microscopic and real-time imaging of tissues without exogenous contrast agents. OCT is a non-contact, non-invasive, micron resolution cross-sectional imaging technique, which is proving its potential in various industrial [1,2] and, biological applications such as ocular disease diagnosis [3], oral cancer [4], breast cancer [5] ovarian cancer [6] and human brain cancer [7], assessment of dental cavities [8], both in ex-vivo and in-vivo [9,10]. Volumetric analysis of normal and cancer breast tissues has been done using support vector machine based texture feature analysis [16]. Certain geometric features in the OCT images, indicative of disease specific morphology, have been identified as diagnostic indicators [3,18,19]. The diversity of diagnostic features, variations in an imaging system, associated calibrations, and, most importantly, difficulty in deriving a consistent and reliable feature base pose difficulty in applying conventional machine learning and pattern recognition techniques

Methods

Results

Conclusion