Saved in:
Bibliographic Details
Main Authors: Boiko, Danylo, Mishkurova, Viktoriia
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.29932
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911728389849088
author Boiko, Danylo
Mishkurova, Viktoriia
author_facet Boiko, Danylo
Mishkurova, Viktoriia
contents Forecasting the progression of neurodegenerative diseases, such as Parkinson's disease, is essential for effective long-term planning and personalized therapeutic intervention. Existing systems typically produce scalar clinical scores that ignore the rich structure of longitudinal neuroimaging, while traditional generative approaches suffer from a loss of anatomical details and blurring subtle progression patterns. To address this, we introduce a novel treatment-conditioned diffusion framework that predicts high-fidelity future brain states by conditioning the generative process on patients' screening DaTscan images and levodopa equivalent daily dose over one year. The pipeline uses a Transformer-based encoder to represent non-linear, time-dependent pharmacological dynamics and optimizes generation through a multi-weight region-of-interest mask that focuses on biologically critical areas. Experimental evaluation shows that our framework maintains sharp anatomical boundaries and significantly improves clinical fidelity relative to the baseline, achieving 14.0% lower MSE, 7.2% lower MAE, and 4.9% higher SSIM.
format Preprint
id arxiv_https___arxiv_org_abs_2605_29932
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Treatment-Conditioned Diffusion for Forecasting Neurodegenerative Disease Progression
Boiko, Danylo
Mishkurova, Viktoriia
Machine Learning
Computer Vision and Pattern Recognition
68T07, 92C55
I.2.m; J.3
Forecasting the progression of neurodegenerative diseases, such as Parkinson's disease, is essential for effective long-term planning and personalized therapeutic intervention. Existing systems typically produce scalar clinical scores that ignore the rich structure of longitudinal neuroimaging, while traditional generative approaches suffer from a loss of anatomical details and blurring subtle progression patterns. To address this, we introduce a novel treatment-conditioned diffusion framework that predicts high-fidelity future brain states by conditioning the generative process on patients' screening DaTscan images and levodopa equivalent daily dose over one year. The pipeline uses a Transformer-based encoder to represent non-linear, time-dependent pharmacological dynamics and optimizes generation through a multi-weight region-of-interest mask that focuses on biologically critical areas. Experimental evaluation shows that our framework maintains sharp anatomical boundaries and significantly improves clinical fidelity relative to the baseline, achieving 14.0% lower MSE, 7.2% lower MAE, and 4.9% higher SSIM.
title Treatment-Conditioned Diffusion for Forecasting Neurodegenerative Disease Progression
topic Machine Learning
Computer Vision and Pattern Recognition
68T07, 92C55
I.2.m; J.3
url https://arxiv.org/abs/2605.29932