HemExp: Clinically-Guided Latent Diffusion for Modeling Hematoma Expansion

2026-06-13

AI Summary

Current approaches struggle to accurately model hematoma expansion after intracerebral hemorrhage with uncertainty awareness, often limited to binary risk stratification or single-volume predictions. This work proposes the first clinically guided latent diffusion model that integrates baseline non-contrast CT scans, clinical variables, and expansion indicators to generate patient-specific follow-up imaging and hemorrhage segmentations. The method introduces a novel controllable latent diffusion mechanism, leveraging a hemorrhage-aware multi-head variational autoencoder and a conditional diffusion model to characterize disease progression through the difference between baseline and follow-up latent representations. This framework enables the generation of spatially resolved hematoma expansion probability maps and supports dynamic perturbation of clinical variables to modulate predictive distributions. Multicenter experiments demonstrate that the model significantly outperforms conventional methods across key clinical metrics, including hematoma volume, intraventricular extension, and mass effect.

Abstract

Hematoma expansion (HE) after spontaneous intracerebral hemorrhage (ICH) is a major determinant of acute triage and treatment decisions in neurosurgical care. However, most existing methods provide either a binary expansion risk or a single follow-up volume, limiting uncertainty-aware decisions. We introduce HemExp, a clinically-guided latent diffusion model that generates patient-specific follow-up non-contrast CT images, along with segmentations of intraparenchymal and intraventricular hemorrhage. Generation is conditioned on baseline imaging, clinical variables, and an explicit expansion indicator, enabling controllable simulation of realistic clinical scenarios. HemExp uses a hemorrhage-aware multi-head variational autoencoder and models progression as the difference between baseline and follow-up latent representations with a conditional diffusion model. The model is trained on paired scans from 450 patients across multiple centers and evaluated on 107 patients from a held-out institution. HemExp produces spatial HE probability maps by generating multiple synthetic follow-up images per patient to estimate distributions of plausible follow-up hematoma volumes. Perturbing clinical inputs such as symptom-onset-to-imaging time or anticoagulant status shifts the predicted follow-up volume distribution. HemExp extends binary predictors and demonstrates robust estimation of clinically relevant outcomes in the imaging space, such as hematoma volume, intraventricular involvement, and mass effects. Overall, our results support controllable latent diffusion as a promising direction for uncertainty-aware modeling of early ICH progression.

Problem

• hematoma expansion
• intracerebral hemorrhage
• uncertainty-aware modeling
• clinical decision support
• non-contrast CT

Innovation

• latent diffusion model
• hematoma expansion
• uncertainty-aware prediction
• clinically-guided generation
• medical image synthesis