Journal
IEEE TRANSACTIONS ON MEDICAL IMAGING
Volume 40, Issue 10, Pages 2748-2758Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TMI.2020.3027665
Keywords
Noise measurement; Training; Photography; Training data; Retinal vessels; Electronic mail; Retinal vessel detection; multi-modal registration; ultra-widefield fundus photography; noisy labels
Categories
Funding
- University of Rochester Research [3B C160189]
- Research to Prevent Blindness
- National Institutes of Health [P30EY001319-35]
Ask authors/readers for more resources
The proposed framework utilizes deep learning and multi-modal registration to achieve accurate vessel detection in UWF fundus photography without requiring de novo labeled UWF FP vessel maps. The method iterates between the registration and weakly supervised learning steps, leveraging UWF FA images for training and progressively improving vessel detection accuracy without the need for manually labeled UWF FP training data.
We propose a deep-learning based annotation-efficient framework for vessel detection in ultra-widefield (UWF) fundus photography (FP) that does not require de novo labeled UWF FP vessel maps. Our approach utilizes concurrently captured UWF fluorescein angiography (FA) images, for which effective deep learning approaches have recently become available, and iterates between a multi-modal registration step and a weakly-supervised learning step. In the registration step, the UWF FA vessel maps detected with a pre-trained deep neural network (DNN) are registered with the UWF FP via parametric chamfer alignment. The warped vessel maps can be used as the tentative training data but inevitably contain incorrect (noisy) labels due to the differences between FA and FP modalities and the errors in the registration. In the learning step, a robust learning method is proposed to train DNNs with noisy labels. The detected FP vessel maps are used for the registration in the following iteration. The registration and the vessel detection benefit from each other and are progressively improved. Once trained, the UWF FP vessel detection DNN from the proposed approach allows FP vessel detection without requiring concurrently captured UWF FA images. We validate the proposed framework on a new UWF FP dataset, PRIME-FP20, and on existing narrow-field FP datasets. Experimental evaluation, using both pixel-wise metrics and the CAL metrics designed to provide better agreement with human assessment, shows that the proposed approach provides accurate vessel detection, without requiring manually labeled UWF FP training data.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available