TY - GEN
T1 - Cross Feature Fusion of Fundus Image and Generated Lesion Map for Referable Diabetic Retinopathy Classification
AU - Mok, Dahyun
AU - Bum, Junghyun
AU - Tai, Le Duc
AU - Choo, Hyunseung
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.
PY - 2025
Y1 - 2025
N2 - Diabetic Retinopathy (DR) is a primary cause of blindness, necessitating early detection and diagnosis. This paper focuses on referable DR classification to enhance the applicability of the proposed method in clinical practice. We develop an advanced cross-learning DR classification method leveraging transfer learning and cross-attention mechanisms. The proposed method employs the Swin U-Net architecture to segment lesion maps from DR fundus images. The Swin U-Net segmentation model, enriched with DR lesion insights, is transferred to generate a lesion map. Both the fundus image and its segmented lesion map are used as complementary inputs for the classification model. A cross-attention mechanism is deployed to improve the model’s ability to capture fine-grained details from the input pairs. Our experiments, utilizing two public datasets, FGADR and EyePACS, demonstrate a superior accuracy of 94.6%, surpassing current state-of-the-art methods by 4.4%. To this end, we aim for the proposed method to be seamlessly integrated into clinical workflows, enhancing accuracy and efficiency in identifying referable DR.
AB - Diabetic Retinopathy (DR) is a primary cause of blindness, necessitating early detection and diagnosis. This paper focuses on referable DR classification to enhance the applicability of the proposed method in clinical practice. We develop an advanced cross-learning DR classification method leveraging transfer learning and cross-attention mechanisms. The proposed method employs the Swin U-Net architecture to segment lesion maps from DR fundus images. The Swin U-Net segmentation model, enriched with DR lesion insights, is transferred to generate a lesion map. Both the fundus image and its segmented lesion map are used as complementary inputs for the classification model. A cross-attention mechanism is deployed to improve the model’s ability to capture fine-grained details from the input pairs. Our experiments, utilizing two public datasets, FGADR and EyePACS, demonstrate a superior accuracy of 94.6%, surpassing current state-of-the-art methods by 4.4%. To this end, we aim for the proposed method to be seamlessly integrated into clinical workflows, enhancing accuracy and efficiency in identifying referable DR.
KW - Cross Fusion
KW - Diabetic Retinopathy
KW - Pseudo Labeling
KW - Referable Classification
KW - Transfer Learning
UR - https://www.scopus.com/pages/publications/85212469245
U2 - 10.1007/978-981-96-0901-7_3
DO - 10.1007/978-981-96-0901-7_3
M3 - Conference contribution
AN - SCOPUS:85212469245
SN - 9789819609000
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 39
EP - 53
BT - Computer Vision – ACCV 2024 - 17th Asian Conference on Computer Vision, Proceedings
A2 - Cho, Minsu
A2 - Laptev, Ivan
A2 - Tran, Du
A2 - Yao, Angela
A2 - Zha, Hongbin
PB - Springer Science and Business Media Deutschland GmbH
T2 - 17th Asian Conference on Computer Vision, ACCV 2024
Y2 - 8 December 2024 through 12 December 2024
ER -