Related Experiment Video
Updated: Jan 16, 2026
Author Spotlight: An Efficient and Robust Software for Automated Fusion of Multiple Preclinical Imaging Modalities
Published on: October 27, 2023
Mixed prototype correction for causal inference in medical image classification
Zhi-Liang Hong1,2,3, Jian-Chuan Yang1,2,3, Xiao-Rui Peng4
1Shengli Clinical Medical College of Fujian Medical University, Fuzhou, China.
Abstract:
The heterogeneity of medical images poses significant challenges to accurate disease diagnosis. To tackle this issue, the impact of such heterogeneity on the causal relationship between image features and diagnostic labels should be incorporated into model design, which however remains under explored. In this paper, we propose a mixed prototype correction for causal inference (MPCCI) method, aimed at mitigating the impact of unseen confounding factors on the causal relationships between medical images and disease labels, so as to enhance the diagnostic accuracy of deep learning models. The MPCCI comprises a causal inference component based on front-door adjustment and an adaptive training strategy. The causal inference component employs a multi-view feature extraction (MVFE) module to establish mediators, and a mixed prototype correction (MPC) module to execute causal interventions. Moreover, the adaptive training strategy incorporates both information purity and maturity metrics to maintain stable model training. Experimental evaluations on four medical image datasets, encompassing CT and ultrasound modalities, demonstrate the superior diagnostic accuracy and reliability of the proposed MPCCI. The code will be available at https://github.com/Yajie-Zhang/MPCCI .
Insights
Medical image heterogeneity challenges diagnosis. We introduce mixed prototype correction for causal inference (MPCCI) to improve deep learning diagnostic accuracy by addressing confounding factors in medical images.
Area of Science:
- Medical imaging and artificial intelligence
- Causal inference in machine learning
- Biomedical data analysis
Background:
- Medical image heterogeneity presents significant challenges for accurate disease diagnosis.
- Current deep learning models often overlook the impact of heterogeneity on causal relationships between image features and diagnostic labels.
- Addressing unseen confounding factors is crucial for reliable medical image diagnosis.
Purpose of the Study:
- To propose a novel method, mixed prototype correction for causal inference (MPCCI), to mitigate the impact of confounding factors in medical images.
- To enhance the diagnostic accuracy and reliability of deep learning models in the presence of medical image heterogeneity.
- To incorporate causal inference principles into deep learning model design for medical image analysis.
Main Methods:
- The proposed MPCCI method integrates a causal inference component using front-door adjustment with an adaptive training strategy.
- A multi-view feature extraction (MVFE) module establishes mediators, while a mixed prototype correction (MPC) module performs causal interventions.
- The adaptive training strategy utilizes information purity and maturity metrics for stable model training.
Main Results:
- Experimental evaluations on four diverse medical image datasets (CT and ultrasound) demonstrated the effectiveness of the MPCCI method.
- The MPCCI approach significantly improved diagnostic accuracy compared to existing methods.
- The proposed method showed superior reliability in handling heterogeneous medical image data.
Conclusions:
- The MPCCI method offers a robust solution for enhancing deep learning-based medical image diagnosis by accounting for heterogeneity and confounding factors.
- Integrating causal inference into deep learning models is a promising direction for improving medical diagnostic systems.
- The developed method has the potential to advance the reliability and accuracy of AI-driven diagnostic tools in healthcare.
