st-DTPM

September 23, 2025 ยท View on GitHub

paper: st-DTPM: Spatial-Temporal Guided Diffusion Transformer Probabilistic Model for Delayed Scan PET Image Prediction

Authors: Ran Hong, Yuxia Huang, Lei Liu, Mengxiao Geng, Zhonghui Wu, Bingxuan Li, Xuemei Wang, Qiegen Liu*

https://ieeexplore.ieee.org/abstract/document/10980366

Date: Apr. 28, 2025

The code and the algorithm are for non-commercial use only.

Copyright 2025, School of Information Engineering, Nanchang University.

Intro :cherry_blossom:

The target of delayed scan PET image prediction is to predict delayed scan PET image from first scan PET image.

target

Motivation :tulip:

The time interval between first and delayed PET image is a crucial factor affecting delayed imaging. And in clinical practice, the time interval for each patient to perform delayed imaging is uncertain.

Proposed :sunflower:

A Diffusion model with Transformer under Spatial-Temporal guidance is proposed. Spatial condition is first scan PET image; Temporal condition is delay time interval.

model

Results :maple_leaf:

result

Training & Testing :evergreen_tree:

**Training for first and delayed PET images. **

--embDTMode and --transEmbDTMode can choose the method of embedding temporal condition into ConvBlock and TransformerBlock, respectively.

Option valueMethod
1each block embedding
2linear cat embedding
3add embedding
4linear add embedding

--condition can choose if use spatial guidance.

--embDT can choose if use temporal guidance.

python runner/train.py --embDTMode=1 --transEmbDTMode=1 --condition=True --embDT=True --runType="train"

Testing for specific delay time interval.

--delayed_time is the delay time interval you given.

python runner/train.py --embDTMode=1 --transEmbDTMode=1 --condition=True --embDT=True --runType="train" --delayed_time=120
Some examples of invertible and variable augmented network: IVNAC, VAN-ICC, iVAN and DTS-INN.

  • Variable Augmented Network for Invertible Modality Synthesis and Fusion [Paper] [Code]

  • Variable augmentation network for invertible MR coil compression [Paper] [Code]

  • Virtual coil augmentation for MR coil extrapoltion via deep learning [Paper] [Code]

  • Synthetic CT Generation via Invertible Network for All-digital Brain PET Attenuation Correction [Paper] [Code]

  • Temporal Image Sequence Separation in Dual-tracer Dynamic PET with an Invertible Network [Paper] [Code]

  • Invertible and Variable Augmented Network for Pretreatment Patient-Specific Quality Assurance Dose Prediction [Paper]

  • A Prior-Guided Joint Diffusion Model in Projection Domain for PET Tracer Conversion [Paper] [Code]

  • Variable augmented neural network for decolorization and multi-exposure fusion [Paper] [Code] [Slide]

  • Positron Emission Tomography Tracer Conversion via Variable Augmented Invertible Network [Paper]