Temporal Representation Learning of Phenotype Trajectories for pCR Prediction in Breast Cancer

2025-09-18

AI Summary

This study addresses the challenge of dynamically predicting pathological complete response (pCR) to neoadjuvant chemotherapy (NACT) in breast cancer patients using longitudinal MRI. We propose a multi-task temporal modeling framework that learns phenotypic evolution trajectories in a latent space, jointly capturing both radiographic appearance changes and temporal continuity to mitigate the high heterogeneity among non-responders. The model processes multi-timepoint MRI features (T₀–T₃), employing trajectory encoding followed by linear classification for early response prediction. On the ISPY-2 dataset, the method achieves a balanced accuracy of 0.761 using only baseline (T₀) data, improves to 0.811 with T₀+T₁, and reaches 0.861 with the full time series (T₀–T₃), significantly outperforming single-timepoint approaches. Our key contribution is the first integration of latent trajectory modeling with multi-task learning for interpretable, robust, and individualized dynamic treatment response prediction.

Technology Category

• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Effective therapy decisions require models that predict the individual response to treatment. This is challenging since the progression of disease and response to treatment vary substantially across patients. Here, we propose to learn a representation of the early dynamics of treatment response from imaging data to predict pathological complete response (pCR) in breast cancer patients undergoing neoadjuvant chemotherapy (NACT). The longitudinal change in magnetic resonance imaging (MRI) data of the breast forms trajectories in the latent space, serving as basis for prediction of successful response. The multi-task model represents appearance, fosters temporal continuity and accounts for the comparably high heterogeneity in the non-responder cohort.In experiments on the publicly available ISPY-2 dataset, a linear classifier in the latent trajectory space achieves a balanced accuracy of 0.761 using only pre-treatment data (T0), 0.811 using early response (T0 + T1), and 0.861 using four imaging time points (T0 -> T3). The code will be made available upon paper acceptance.

Problem

• Predicting breast cancer treatment response using imaging trajectories
• Modeling longitudinal MRI changes for pCR prediction in NACT
• Addressing patient heterogeneity in non-responder cohort representation

Innovation

• Learning phenotype trajectories from imaging data
• Multi-task model for temporal continuity and heterogeneity
• Latent trajectory space for pCR prediction accuracy