eduzhai > Applied Sciences > Engineering >

Exploiting the Transferability of Deep Learning Systems Across Multi-modal Retinal Scans for Extracting Retinopathy Lesions

  • Save

... pages left unread,continue reading

Document pages: 5 pages

Abstract: Retinal lesions play a vital role in the accurate classification of retinalabnormalities. Many researchers have proposed deep lesion-aware screeningsystems that analyze and grade the progression of retinopathy. However, to thebest of our knowledge, no literature exploits the tendency of these systems togeneralize across multiple scanner specifications and multi-modal imagery.Towards this end, this paper presents a detailed evaluation of semanticsegmentation, scene parsing and hybrid deep learning systems for extracting theretinal lesions such as intra-retinal fluid, sub-retinal fluid, hard exudates,drusen, and other chorioretinal anomalies from fused fundus and opticalcoherence tomography (OCT) imagery. Furthermore, we present a novel strategyexploiting the transferability of these models across multiple retinal scannerspecifications. A total of 363 fundus and 173,915 OCT scans from seven publiclyavailable datasets were used in this research (from which 297 fundus and 59,593OCT scans were used for testing purposes). Overall, a hybrid retinal analysisand grading network (RAGNet), backboned through ResNet-50, stood first forextracting the retinal lesions, achieving a mean dice coefficient score of0.822. Moreover, the complete source code and its documentation are releasedat: this http URL.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×