Logo image
A multiple-instance learning framework for diabetic retinopathy screening
Journal article   Peer reviewed

A multiple-instance learning framework for diabetic retinopathy screening

Gwénolé Quellec, Mathieu Lamard, Michael D Abràmoff, Etienne Decencière, Bruno Lay, Ali Erginay, Béatrice Cochener and Guy Cazuguel
Medical Image Analysis, Vol.16(6), pp.1228-1240
08/2012
DOI: 10.1016/j.media.2012.06.003
PMID: 22850462

View Online

Abstract

[Display omitted] ► Lesion detection is supervised without manual segmentation. ► Large image datasets can be used for training. ► A large diabetic retinopathy screening dataset (>100,000 images) is presented. ► All eight types of diabetic retinopathy lesions are detected. ► Diabetic retinopathy is detected with high sensitivity/specificity. A novel multiple-instance learning framework, for automated image classification, is presented in this paper. Given reference images marked by clinicians as relevant or irrelevant, the image classifier is trained to detect patterns, of arbitrary size, that only appear in relevant images. After training, similar patterns are sought in new images in order to classify them as either relevant or irrelevant images. Therefore, no manual segmentations are required. As a consequence, large image datasets are available for training. The proposed framework was applied to diabetic retinopathy screening in 2-D retinal image datasets: Messidor (1200 images) and e-ophtha, a dataset of 25,702 examination records from the Ophdiat screening network (107,799 images). In this application, an image (or an examination record) is relevant if the patient should be referred to an ophthalmologist. Trained on one half of Messidor, the classifier achieved high performance on the other half of Messidor (Az=0.881) and on e-ophtha (Az=0.761). We observed, in a subset of 273 manually segmented images from e-ophtha, that all eight types of diabetic retinopathy lesions are detected.
Diabetic Retinopathy Multiple-instance learning Lesion detection Pathology screening

Details

Logo image