Polyp-D2ATL: Deep Domain-Adaptive Transfer Learning for Colorectal Polyp Classification under Label Distribution Shift

2026-06-12

AI Summary

This study addresses the challenges of label distribution shift, data imbalance, and cross-modal discrepancies that hinder generalization in automatic colorectal polyp classification. To tackle these issues, the work systematically introduces deep domain adaptation and transfer learning into the NICE classification framework for the first time, proposing a novel approach that integrates task-specific training strategies with multi-metric optimization. The proposed method substantially enhances model robustness and generalization under distribution shifts and cross-domain settings. On the PICCOLO validation set, it achieves 82.38% accuracy, 77.49% Macro-F1 score, and 87.47% specificity, with consistent performance improvements also observed on the test set.

Abstract

Early and highly accurate prediction of colorectal polyps, as an important sign of one of the most dangerous types of cancer, will result in saving more lives. Despite the advancements in colorectal polyp classification, many challenges remain in obtaining an automated polyp prediction system that is able to diagnose the difficult-to-predict polyps accompanied by different features in real scenarios, where the model can handle imbalanced data, label distribution shift, and cross-modality generalization successfully. In this study, we propose Polyp-D2ATL, a novel framework accompanied by a specific training strategy, which mitigates these limitations and effectively predicts the different classes of polyps belonging to the NICE classification. Our extensive experiments on the PICCOLO validation and test sets demonstrate that the proposed Polyp-D2ATL significantly outperforms existing state-of-the-art models across various reliable metrics, achieving an accuracy of 82.38%, a Macro-F1 of 77.49%, and a specificity of 87.47% on the validation set, alongside consistent improvements on the held-out test set which demonstrates the generalization capacity and clinical applicability of the proposed approach.

Problem

• colorectal polyp classification
• label distribution shift
• imbalanced data
• cross-modality generalization

Innovation

• Domain-Adaptive Transfer Learning
• Label Distribution Shift
• Colorectal Polyp Classification
• Imbalanced Data
• Cross-Modality Generalization