Causality matters in medical imaging

Abstract

This article discusses how the language of causality can shed new light on the major challenges in machine learning for medical imaging: 1) data scarcity, which is the limited availability of high-quality annotations, and 2) data mismatch, whereby a trained algorithm may fail to generalize in clinical practice. Looking at these challenges through the lens of causality allows decisions about data collection, annotation procedures, and learning strategies to be made (and scrutinized) more transparently. We discuss how causal relationships between images and annotations can not only have profound effects on the performance of predictive models, but may even dictate which learning strategies should be considered in the first place. For example, we conclude that semi-supervision may be unsuitable for image segmentation---one of the possibly surprising insights from our causal analysis, which is illustrated with representative real-world examples of computer-aided diagnosis (skin lesion classification in dermatology) and radiotherapy (automated contouring of tumours). We highlight that being aware of and accounting for the causal relationships in medical imaging data is important for the safe development of machine learning and essential for regulation and responsible reporting. To facilitate this we provide step-by-step recommendations for future studies.

20 pages, 5 figures, 4 tables

References in corpus (6)

• Model Cards for Model Reporting
• Why rankings of biomedical image analysis competitions should be interpreted with care
• External Validity: From Do-Calculus to Transportability Across Populations
• SynSeg-Net: Synthetic Segmentation Without Target Modality Ground Truth
• Inferring deterministic causal relations
• A Causal Bayesian Networks Viewpoint on Fairness

Cited by in corpus (19)

• Explainable artificial intelligence (XAI) in deep learning-based medical image analysis
• The Liver Tumor Segmentation Benchmark (LiTS)
• Artificial Intelligence for Digital and Computational Pathology
• Learning Disentangled Representations in the Imaging Domain
• Data synthesis and adversarial networks: A review and meta-analysis in cancer imaging
• Risk of Bias in Chest Radiography Deep Learning Foundation Models
• Responsible and Regulatory Conform Machine Learning for Medicine: A Survey of Challenges and Solutions
• Applications of statistical causal inference in software engineering
• Fairness in Agreement With European Values: An Interdisciplinary Perspective on AI Regulation
• Image-level Harmonization of Multi-Site Data using Image-and-Spatial Transformer Networks
• RadEdit: stress-testing biomedical vision models via diffusion image editing
• Robustness and Cybersecurity in the EU Artificial Intelligence Act
• Advances in Medical Image Segmentation: A Comprehensive Survey with a Focus on Lumbar Spine Applications
• Advances in Automated Fetal Brain MRI Segmentation and Biometry: Insights from the FeTA 2024 Challenge
• Disentangling representations of retinal images with generative models
• A Systematic Review on the Generative AI Applications in Human Medical Genomics
• Sharing Generative Models Instead of Private Data: A Simulation Study on Mammography Patch Classification
• polyDAG: Polynomial Acyclicity Constraints for Efficient Continuous Causal Discovery in Visual Semantic Graphs
• Requirement analysis for an artificial intelligence model for the diagnosis of the COVID-19 from chest X-ray data