University of British Columbia
Diabetic retinopathy classification using an efficient convolutional neural network
Abstract
dc:descriptionDiabetic Retinopathy (DR) is a diabetic complication that affects the eyes and may lead to blurred vision or even blindness. The diagnosis of DR through retinal fundus images is traditionally performed by ophthalmologists who inspect for the presence and significance of many subtle features, a process which is cumbersome and time-consuming. As there are many undiagnosed and untreated cases of DR, DR screening of all diabetic patients is a huge challenge. Deep convolutional neural network (CNN) has rapidly become a powerful tool for analyzing medical images. There have been previous works which use deep learning models to detect DR automatically. However, these methods employed very deep CNNs which require vast computational resources. Thus, there is a need for more computationally efficient deep learning models for automatic DR diagnosis. The primary objective of this research is to develop a robust and computationally efficient deep learning model to diagnose DR automatically. In the first part of this thesis, we propose a computationally efficient deep CNN model MobileNet-Dense which is based on the recently proposed MobileNetV2 and DenseNet models. The effectiveness of the proposed MobileNet-Dense model is demonstrated using two widely used benchmark datasets, CIFAR-10 and CIFAR-100. In the second part of the thesis, we propose an automatic DR classification system based on the ensemble of the proposed MobileNet-Dense model and the MobileNetV2 model. The performance of our system is evaluated and compared with some of the state-of-the-art methods using two independent DR datasets, the EyePACS dataset and the Messidor database. On the EyePACS dataset, our system achieves a quadratic weighted kappa (QWK) score of 0.852 compared to a QWK score of 0.849 achieved by the benchmark method while using 32% fewer parameters and 73% fewer multiply-adds (MAdds). On the Messidor database, our system outperforms the state-of-the-art method on both Normal/Abnormal and Referable/Non-Referable classification tasks.
Degree
thesis:*- Name thesis:degree_name
- Master of Applied Science - MASc
- Level thesis:degree_level
- master's
- Discipline thesis:degree_discipline
- Electrical and Computer Engineering
- Grantor dc:publisher
- University of British Columbia
- Year dc:date
- 2019
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Gao, Jiaxi
Rights
dc:rights- Statement dc:rights
-
- Attribution-NonCommercial-NoDerivatives 4.0 International
- Language dc:language
- eng
Identifiers
dc:identifier.*- Handle dc:identifier
- http://hdl.handle.net/2429/70037