One of the core challenges in applying machine learning and artificial intelligence to medicine is the limited availability of annotated medical data. Unlike in other applications of machine learning, where an abundance of labeled data is available, the labeling and annotation of medical data and images require a major effort of manual work by expert clinicians who do not have the time to annotate manually. In this work, we propose a new deep learning technique (SLIVER-net), to predict clinical features from 3-dimensional volumes using a limited number of manually annotated examples. SLIVER-net is based on transfer learning, where we borrow information about the structure and parameters of the network from publicly available large datasets. Since public volume data are scarce, we use 2D images and account for the 3-dimensional structure using a novel deep learning method which tiles the volume scans, and then adds layers that leverage the 3D structure. In order to illustrate its utility, we apply SLIVER-net to predict risk factors for progression of age-related macular degeneration (AMD), a leading cause of blindness, from optical coherence tomography (OCT) volumes acquired from multiple sites. SLIVER-net successfully predicts these factors despite being trained with a relatively small number of annotated volumes (hundreds) and only dozens of positive training examples. Our empirical evaluation demonstrates that SLIVER-net significantly outperforms standard state-of-the-art deep learning techniques used for medical volumes, and its performance is generalizable as it was validated on an external testing set. In a direct comparison with a clinician panel, we find that SLIVER-net also outperforms junior specialists, and identifies AMD progression risk factors similarly to expert retina specialists.

PURPOSE: To assess the relationship between qualitative diabetic retinopathy (DR) scales with the precise numbers and surface area of DR lesions within the Early Treatment Diabetic Retinopathy Study (ETDRS) standard seven field (S7F) region on ultrawide-field (UWF) color fundus images. METHODS: In this study, we collected UWF images from adult patients with diabetes. Poor-quality images and eyes with any pathology precluding assessment of DR severity were excluded. The DR lesions were manually segmented. DR severity was graded according to the International Clinical Diabetic Retinopathy (ICDR) and AA protocol by two masked graders within the ETDRS S7F. These lesions numbers and surface area were computed and correlated against the DR scores using the Kruskal-Wallis H test. Cohens Kappa was performed to determine the agreement between two graders. RESULTS: One thousand five hundred and twenty eyes of 869 patients (294 females, 756 right eyes) with a mean age of 58.7 years were included. 47.4% were graded as no DR, 2.2% as mild non-proliferative DR (NPDR), 24.0% as moderate NPDR, 6.3% as severe NPDR, and 20.1% as proliferative DR (PDR). The area and number of DR lesions generally increased as the ICDR level increased up to severe NPDR, but decreased from severe NPDR to PDR. There was perfect intergrader agreement on the DR severity. CONCLUSION: A quantitative approach reveals that DR lesions number and area generally correlate with ICDR-based categorical DR severity levels with an increasing trend in the number and area of DR lesions from mild to severe NPDR and a decrease from severe NPDR to PDR.