Team Westwood Solution for MIDOG 2025 Challenge

2025-08-29

AI Summary

This work addresses two key tasks in the MIDOG 2025 challenge: cross-domain mitosis detection and fine-grained classification of atypical mitoses. To tackle domain shift and improve robustness, we propose a multi-stage collaborative framework: (1) initial high-sensitivity candidate region detection using nnUNetV2; (2) multi-scale feature extraction via ensemble convolutional backbones—EfficientNet-B3/B5, EfficientNetV2-S, and InceptionV3; and (3) robust classification using a Random Forest ensemble. Our approach effectively mitigates domain bias while preserving discriminative capacity for rare pathological patterns. On the preliminary test set, it achieves an F1-score of 0.7450 for mitosis detection and a balanced accuracy of 0.8722 for atypical mitosis classification. The framework delivers a generalizable, end-to-end solution for automated identification of rare abnormal mitotic events in digital pathology images, advancing clinical-grade computational pathology tools.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Computer Vision: Adversarial Attacks & Robustness
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Search and Retrieval-Augmented AI: Vertical and domain-specific search

Abstract

This abstract presents our solution (Team Westwood) for mitosis detection and atypical mitosis classification in the MItosis DOmain Generalization (MIDOG) 2025 challenge. For mitosis detection, we trained an nnUNetV2 for initial mitosis candidate screening with high sensitivity, followed by a random forest classifier ensembling predictions of three convolutional neural networks (CNNs): EfficientNet-b3, EfficientNet-b5, and EfficientNetV2-s. For the atypical mitosis classification, we trained another random forest classifier ensembling the predictions of three CNNs: EfficientNet-b3, EfficientNet-b5, and InceptionV3. On the preliminary test set, our solution achieved an F1 score of 0.7450 for track 1 mitosis detection, and a balanced accuracy of 0.8722 for track 2 atypical mitosis classification.

Problem

• Detecting mitosis in histopathology images
• Classifying atypical mitotic figures accurately
• Generalizing across diverse domain datasets

Innovation

• nnUNetV2 for initial mitosis candidate screening
• Random forest classifier ensembling three CNN predictions
• EfficientNet and InceptionV3 models for atypical mitosis classification