[vc_empty_space][vc_empty_space]
Diabetic Retinopathy Classification Using A Hybrid and Efficient MobileNetV2-SVM Model
Taufiqurrahman S.a, Handayani A.a, Hermanto B.R.a, Mengko T.L.E.R.a
a Institut Teknologi Bandung, Biomedical Engineering Program, School of Electrical Engineering and Informatics, Bandung, Indonesia
[vc_row][vc_column][vc_row_inner][vc_column_inner][vc_separator css=”.vc_custom_1624529070653{padding-top: 30px !important;padding-bottom: 30px !important;}”][/vc_column_inner][/vc_row_inner][vc_row_inner layout=”boxed”][vc_column_inner width=”3/4″ css=”.vc_custom_1624695412187{border-right-width: 1px !important;border-right-color: #dddddd !important;border-right-style: solid !important;border-radius: 1px !important;}”][vc_empty_space][megatron_heading title=”Abstract” size=”size-sm” text_align=”text-left”][vc_column_text]© 2020 IEEE.Deep learning has been proposed as one of the automated solutions for diabetic retinopathy (DR) severity classification problem. However, most of the successful deep learning models are based on large convolutional neural network (CNN) architectures, requiring a vast volume of training data as well as dedicated computational resources. In this study, we used MobileNetV2 architecture, which was considered a small-scale architecture (4.2 million trainable parameters), to perform DR classification task in APTOS 2019 dataset (3662 color retinal images). We used the generic MobileNetV2 pre-trained weights from ImageNet as initialization and implemented data augmentation and resampling during training. We further optimized our model by combining it with an SVM classifier, resulting a hybrid and computationally efficient deep learning model, MobileNetV2-SVM. This model obtained quadratic weighted kappa score of 0.925, 85% accuracy, and area under receiving operating characteristic (AUROC) of 1.00, 0.82, 0.94, 0.94, 0.93 for normal, mild, moderate, severe, and proliferative DR classes, respectively; which is better or at least comparable with the larger architecture performance on the same dataset. Our result shows that with proper optimization strategy, a relatively small and generic CNN architecture, can achieve promising DR classification performance, and even outperform the performance of CNN model with larger architecture.[/vc_column_text][vc_empty_space][vc_separator css=”.vc_custom_1624528584150{padding-top: 25px !important;padding-bottom: 25px !important;}”][vc_empty_space][megatron_heading title=”Author keywords” size=”size-sm” text_align=”text-left”][vc_column_text]Classification performance,Classification tasks,Color retinal images,Computational resources,Computationally efficient,Diabetic retinopathy,Optimization strategy,Receiving operating characteristics[/vc_column_text][vc_empty_space][vc_separator css=”.vc_custom_1624528584150{padding-top: 25px !important;padding-bottom: 25px !important;}”][vc_empty_space][megatron_heading title=”Indexed keywords” size=”size-sm” text_align=”text-left”][vc_column_text]APTOS 2019,Classification,Convolutional neural network,Deep learning,Diabetic retinopathy[/vc_column_text][vc_empty_space][vc_separator css=”.vc_custom_1624528584150{padding-top: 25px !important;padding-bottom: 25px !important;}”][vc_empty_space][megatron_heading title=”Funding details” size=”size-sm” text_align=”text-left”][vc_column_text][/vc_column_text][vc_empty_space][vc_separator css=”.vc_custom_1624528584150{padding-top: 25px !important;padding-bottom: 25px !important;}”][vc_empty_space][megatron_heading title=”DOI” size=”size-sm” text_align=”text-left”][vc_column_text]https://doi.org/10.1109/TENCON50793.2020.9293739[/vc_column_text][/vc_column_inner][vc_column_inner width=”1/4″][vc_column_text]Widget Plumx[/vc_column_text][/vc_column_inner][/vc_row_inner][/vc_column][/vc_row][vc_row][vc_column][vc_separator css=”.vc_custom_1624528584150{padding-top: 25px !important;padding-bottom: 25px !important;}”][/vc_column][/vc_row]