Problem
Research questions and friction points this paper is trying to address.
Physiological time series
Signal restoration
Cyclostationarity
Diffusion models
Cyclostationary Phase Conditioning for Medical Time Series Diffusion
🤖 AI Summary
This study addresses the challenge of high-fidelity diagnostic signal recovery in physiological time series, where noise interference frequently causes the loss of morphological features. To overcome this, we propose a diffusion model based on cyclostationary phase conditioning that explicitly encodes periodic structures through translation-covariant wavelet representations and dense phase conditions. Additionally, a training-free cyclostationarity index is designed to characterize signal properties, and a dual-coupled sampling strategy is developed to accelerate the generation process. This work significantly enhances the recovery quality of multimodal physiological signals while reducing the number of network evaluations fivefold, thereby achieving an effective balance between high precision and computational efficiency.
📝 Abstract
Many physiological time series, such as cardiac and brain recordings, exhibit cyclostationarity: their statistics vary periodically with an underlying cycle phase. Corruption from motion, poor contact, and physiological interference obscures morphology needed for diagnosis, making signal restoration essential. Existing diffusion approaches condition on corrupted observations alone and must learn cyclic structure implicitly. We instead propose two inductive biases which encode cyclostationarity: a shift-covariant wavelet representation and dense per-sample phase conditioning inferred from the corrupted input. We further introduce a training-free cyclostationarity index that quantifies phase structure and predicts when phase conditioning will help. Finally, we propose antithetic coupling of reverse trajectories to reduce sampling variance while achieving comparable performance with fivefold fewer network evaluations. Across modalities, our results show that explicitly encoding measurable cyclic structure improves physiological time-series restoration.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Cyclostationary Phase Conditioning
Diffusion Models
Wavelet Representation
Antithetic Coupling
Medical Time Series
🔎 Similar Papers
No similar papers found.
