Simulating Post-Neoadjuvant Chemotherapy Breast Cancer MRI via Diffusion Model with Prompt Tuning

2025-09-28

AI Summary

This study addresses the challenge of predicting neoadjuvant chemotherapy (NAC) response in breast cancer. We propose a diffusion-based conditional generation method that synthesizes post-NAC dynamic contrast-enhanced MRI (DCE-MRI) at 3- or 12-week timepoints from pre-treatment scans. Methodologically, we integrate maximum intensity projection (MIP) representations with a clinical prompt tuning mechanism: key prognostic biomarkers—including hormone receptor status and Ki-67 index—are encoded as learnable prompt vectors to steer the model toward pathologic complete response (pCR)-associated morphological features. Quantitative evaluation demonstrates significant improvements over GAN- and VAE-based baselines in FID, LPIPS, and tumor volume change consistency. Ablation studies confirm that prompt tuning enhances sensitivity to treatment response. Our approach establishes an interpretable, clinically translatable paradigm for personalized NAC response assessment and precision decision support.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Generation
• Machine Learning: Deep Generative Models & Autoencoders

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
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Neoadjuvant chemotherapy (NAC) is a common therapy option before the main surgery for breast cancer. Response to NAC is monitored using follow-up dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). Accurate prediction of NAC response helps with treatment planning. Here, we adopt maximum intensity projection images from DCE-MRI to generate post-treatment images (i.e., 3 or 12 weeks after NAC) from pre-treatment images leveraging the emerging diffusion model. We introduce prompt tuning to account for the known clinical factors affecting response to NAC. Our model performed better than other generative models in image quality metrics. Our model was better at generating images that reflected changes in tumor size according to pCR compared to other models. Ablation study confirmed the design choices of our method. Our study has the potential to help with precision medicine.

Problem

• Predicting post-NAC breast cancer MRI images from pre-treatment scans
• Generating accurate tumor size changes using diffusion models
• Incorporating clinical factors via prompt tuning for precision medicine

Innovation

• Using diffusion model for MRI image generation
• Applying prompt tuning for clinical factors
• Generating post-treatment images from pre-treatment data