Deep structure tensor graph search framework for automated extraction and characterization of retinal layers and fluid pathology in retinal SD-OCT scans

• Published in: Computers in biology and medicine Vol. 105; pp. 112 - 124
• Main Authors: Hassan, Taimur, Akram, Muhammad Usman, Masood, Muhammad Furqan, Yasin, Ubaidullah
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.02.2019; Elsevier Limited
• Subjects: Accuracy; Annotations; Armed forces; Artificial neural networks; Automation; Computational fluid dynamics; Convolutional neural network (CNN); Cysts; Datasets; Diabetes; Diabetic retinopathy; Diagnostic systems; Edema; Graph search; Humans; Image Interpretation, Computer-Assisted; Internal Medicine; Macular degeneration; Macular Degeneration - diagnostic imaging; Maculopathy; Medical imaging; Neural networks; Neural Networks, Computer; Ophthalmology; Optical Coherence Tomography; Optical coherence tomography (OCT); Other; Pathology; People with disabilities; Retina; Retina - diagnostic imaging; Segmentation; Tensors; Tomography, Optical Coherence; Visualization; Maculopathy; Ophthalmology; Convolutional neural network (CNN); Optical coherence tomography (OCT); Graph search
• ISSN: 0010-4825; 1879-0534; 1879-0534
• DOI: 10.1016/j.compbiomed.2018.12.015

Maculopathy is a group of retinal disorders that affect macula and cause severe visual impairment if not treated in time. Many computer-aided diagnostic methods have been proposed over the past that automatically detect macular diseases. However, to our best knowledge, no literature is available that provides an end-to-end solution for analyzing healthy and diseased macular pathology. This paper proposes a vendor-independent deep convolutional neural network and structure tensor graph search-based segmentation framework (CNN-STGS) for the extraction and characterization of retinal layers and fluid pathology, along with 3-D retinal profiling. CNN-STGS works by first extracting nine layers from an optical coherence tomography (OCT) scan. Afterward, the extracted layers, combined with a deep CNN model, are used to automatically segment cyst and serous pathology, followed by the autonomous 3-D retinal profiling. CNN-STGS has been validated on publicly available Duke datasets (containing a cumulative of 42,281 scans from 439 subjects) and Armed Forces Institute of Ophthalmology dataset (containing 4260 OCT scans of 51 subjects), which are acquired through different OCT machinery. The performance of the CNN-STGS framework is validated through the marked annotations, and it significantly outperforms the existing solutions in various metrics. The proposed CNN-STGS framework achieved a mean Dice coefficient of 0.906 for segmenting retinal fluids, along with an accuracy of 98.75% for characterizing cyst and serous fluid from diseased retinal OCT scans. •This paper presents a vendor-independent framework for extracting retinal information.•CNN-STGS is validated on 46,541 OCT scans from four publicly available datasets.•CNN-STGS is invariant to scan quality, acquisition machinery, and eye pathology.•It can pick even the slightest fluid variation and low-intensity layer information.•CNN-STGS significantly outperforms state-of-the-art solutions in various metrics.