ConvNeXt with Histopathology-Specific Augmentations for Mitotic Figure Classification

2025-08-29

AI Summary

Accurately distinguishing atypical mitotic figures (AMFs) from normal mitotic figures (NMFs) is critical for cancer grading and prognostic assessment, yet remains challenging due to subtle morphological differences, high intra-class variability, multi-center domain shift, scarce expert annotations, and severe class imbalance. To address these challenges, we propose a lightweight, pathology-oriented classification framework built upon a ConvNeXt backbone. Our method incorporates stain-specific augmentation and elastic deformation for robust feature learning, integrates balanced sampling with multi-center data fusion, and employs grouped cross-validation to mitigate domain shift. Evaluated on the MIDOG 2025 challenge leaderboard, our approach achieves a balanced accuracy of 0.8961—ranking first—and demonstrates substantial improvements in generalizability, robustness, and clinical deployability across diverse histopathological domains.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Computer Vision: Adversarial Attacks & Robustness
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Accurate mitotic figure classification is crucial in computational pathology, as mitotic activity informs cancer grading and patient prognosis. Distinguishing atypical mitotic figures (AMFs), which indicate higher tumor aggressiveness, from normal mitotic figures (NMFs) remains challenging due to subtle morphological differences and high intra-class variability. This task is further complicated by domain shifts, including variations in organ, tissue type, and scanner, as well as limited annotations and severe class imbalance. To address these challenges in Track 2 of the MIDOG 2025 Challenge, we propose a solution based on the lightweight ConvNeXt architecture, trained on all available datasets (AMi-Br, AtNorM-Br, AtNorM-MD, and OMG-Octo) to maximize domain coverage. Robustness is enhanced through a histopathology-specific augmentation pipeline, including elastic and stain-specific transformations, and balanced sampling to mitigate class imbalance. A grouped 5-fold cross-validation strategy ensures reliable evaluation. On the preliminary leaderboard, our model achieved a balanced accuracy of 0.8961, ranking among the top entries. These results highlight that broad domain exposure combined with targeted augmentation strategies is key to building accurate and generalizable mitotic figure classifiers.

Problem

• Distinguishing atypical from normal mitotic figures due to subtle morphological differences
• Addressing domain shifts across organ, tissue type, and scanner variations
• Overcoming limited annotations and severe class imbalance in classification

Innovation

• ConvNeXt architecture for lightweight classification
• Histopathology-specific augmentation pipeline for robustness
• Grouped cross-validation for reliable evaluation