Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Agent Aggregator with Mask Denoise Mechanism for Histopathology Whole Slide Image Analysis

Xitong Ling, Minxi Ouyang, Yizhi Wang, Xinrui Chen, Renao Yan, Hongbo Chu, Junru Cheng, Tian Guan, Sufang Tian, Xiaoping Liu, Yonghong He

Abstract

Histopathology analysis is the gold standard for medical diagnosis. Accurate classification of whole slide images (WSIs) and region-of-interests (ROIs) localization can assist pathologists in diagnosis. The gigapixel resolution of WSI and the absence of fine-grained annotations make direct classification and analysis challenging. In weakly supervised learning, multiple instance learning (MIL) presents a promising approach for WSI classification. The prevailing strategy is to use attention mechanisms to measure instance importance for classification. However, attention mechanisms fail to capture inter-instance information, and self-attention causes quadratic computational complexity. To address these challenges, we propose AMD-MIL, an agent aggregator with a mask denoise mechanism. The agent token acts as an intermediate variable between the query and key for computing instance importance. Mask and denoising matrices, mapped from agents-aggregated value, dynamically mask low-contribution representations and eliminate noise. AMD-MIL achieves better attention allocation by adjusting feature representations, capturing micro-metastases in cancer, and improving interpretability. Extensive experiments on CAMELYON-16, CAMELYON-17, TCGA-KIDNEY, and TCGA-LUNG show AMD-MIL's superiority over state-of-the-art methods.

Agent Aggregator with Mask Denoise Mechanism for Histopathology Whole Slide Image Analysis

Abstract

Histopathology analysis is the gold standard for medical diagnosis. Accurate classification of whole slide images (WSIs) and region-of-interests (ROIs) localization can assist pathologists in diagnosis. The gigapixel resolution of WSI and the absence of fine-grained annotations make direct classification and analysis challenging. In weakly supervised learning, multiple instance learning (MIL) presents a promising approach for WSI classification. The prevailing strategy is to use attention mechanisms to measure instance importance for classification. However, attention mechanisms fail to capture inter-instance information, and self-attention causes quadratic computational complexity. To address these challenges, we propose AMD-MIL, an agent aggregator with a mask denoise mechanism. The agent token acts as an intermediate variable between the query and key for computing instance importance. Mask and denoising matrices, mapped from agents-aggregated value, dynamically mask low-contribution representations and eliminate noise. AMD-MIL achieves better attention allocation by adjusting feature representations, capturing micro-metastases in cancer, and improving interpretability. Extensive experiments on CAMELYON-16, CAMELYON-17, TCGA-KIDNEY, and TCGA-LUNG show AMD-MIL's superiority over state-of-the-art methods.
Paper Structure (14 sections, 13 equations, 7 figures, 3 tables, 1 algorithm)

This paper contains 14 sections, 13 equations, 7 figures, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Multiple Instance Learning for WSI Analysis
  4. Approximate Self-Attention Mechanism
  5. Methodology
  6. MIL and Feature extraction
  7. Attention Aggregator
  8. Agent Mask Denoise Mechanism
  9. Experiments
  10. Datasets and Evaluation Metrics
  11. Implementation Details
  12. Comparison with State-of-the-Art Methods
  13. Interpretability Analysis
  14. Ablation Study

Figures (7)

  • Figure 1: Comparison of core modules: (a) pooling agents. (b) proposed trainable agents. (c) self-attention mechanism. (d) proposed agent aggregator with mask denoise mechanism. Mask and denoising are learnable matrices.
  • Figure 2: Overall process: (a) the preprocess of WSI. (b) overall framework of AMD-MIL. (c) proposed mask denoise mechanism.
  • Figure 3: Visualization of AMD-MIL Attention Distribution Compared to Official Annotations on CAMELYON dataset.
  • Figure 4: Attention distribution of different agent tokens.
  • Figure 5: Influence of the number of agent tokens
  • ...and 2 more figures