An Intelligent Grading Model for Myopic Maculopathy Based on Long-Tailed Learning DOI Creative Commons
Bo Zheng, Wang Chen,

M M Zhang

et al.

Translational Vision Science & Technology, Journal Year: 2025, Volume and Issue: 14(3), P. 4 - 4

Published: March 6, 2025

To develop an intelligent grading model for myopic maculopathy based on a long-tail learning framework, using the improved loss function LTBSoftmax. The addresses distribution problem in data to provide preliminary grading, aiming improve capability and efficiency. This study includes set of 7529 color fundus photographs. Experienced ophthalmologists meticulously annotated ground truth. A new was constructed LTBSoftmax, which predicts lesions by locally enhancing feature extraction with ND Block. Standard metrics were selected evaluate LTBSoftmax model. demonstrated excellent performance diagnosing four types maculopathy, achieving κ coefficient 88.89%. Furthermore, model's size is 18.7 MB, relatively smaller compared traditional models, indicating that not only achieves high level agreement expert diagnoses but also more efficient terms both storage computational resources. These further validate well-conceived design superiority practical applications. system, long-tailed strategies, effectively improves classification offering tool clinicians, particularly areas limited translates research into maculopathy. It imbalance function, accuracy By Block, it provides reliable support especially resource-limited settings.

Language: Английский

An Intelligent Grading Model for Myopic Maculopathy Based on Long-Tailed Learning DOI Creative Commons
Bo Zheng, Wang Chen,

M M Zhang

et al.

Translational Vision Science & Technology, Journal Year: 2025, Volume and Issue: 14(3), P. 4 - 4

Published: March 6, 2025

To develop an intelligent grading model for myopic maculopathy based on a long-tail learning framework, using the improved loss function LTBSoftmax. The addresses distribution problem in data to provide preliminary grading, aiming improve capability and efficiency. This study includes set of 7529 color fundus photographs. Experienced ophthalmologists meticulously annotated ground truth. A new was constructed LTBSoftmax, which predicts lesions by locally enhancing feature extraction with ND Block. Standard metrics were selected evaluate LTBSoftmax model. demonstrated excellent performance diagnosing four types maculopathy, achieving κ coefficient 88.89%. Furthermore, model's size is 18.7 MB, relatively smaller compared traditional models, indicating that not only achieves high level agreement expert diagnoses but also more efficient terms both storage computational resources. These further validate well-conceived design superiority practical applications. system, long-tailed strategies, effectively improves classification offering tool clinicians, particularly areas limited translates research into maculopathy. It imbalance function, accuracy By Block, it provides reliable support especially resource-limited settings.

Language: Английский

Citations

0