Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

PASTA: Pathology-Aware MRI to PET Cross-Modal Translation with Diffusion Models

Yitong Li, Igor Yakushev, Dennis M. Hedderich, Christian Wachinger

TL;DR

The qualitative and quantitative results confirm that the synthesized PET scans from PASTA not only reach the best quantitative scores but also preserve the pathology correctly, almost reaching the performance of actual PET.

Abstract

Positron emission tomography (PET) is a well-established functional imaging technique for diagnosing brain disorders. However, PET's high costs and radiation exposure limit its widespread use. In contrast, magnetic resonance imaging (MRI) does not have these limitations. Although it also captures neurodegenerative changes, MRI is a less sensitive diagnostic tool than PET. To close this gap, we aim to generate synthetic PET from MRI. Herewith, we introduce PASTA, a novel pathology-aware image translation framework based on conditional diffusion models. Compared to the state-of-the-art methods, PASTA excels in preserving both structural and pathological details in the target modality, which is achieved through its highly interactive dual-arm architecture and multi-modal condition integration. A cycle exchange consistency and volumetric generation strategy elevate PASTA's capability to produce high-quality 3D PET scans. Our qualitative and quantitative results confirm that the synthesized PET scans from PASTA not only reach the best quantitative scores but also preserve the pathology correctly. For Alzheimer's classification, the performance of synthesized scans improves over MRI by 4%, almost reaching the performance of actual PET. Code is available at https://github.com/ai-med/PASTA.

PASTA: Pathology-Aware MRI to PET Cross-Modal Translation with Diffusion Models

TL;DR

The qualitative and quantitative results confirm that the synthesized PET scans from PASTA not only reach the best quantitative scores but also preserve the pathology correctly, almost reaching the performance of actual PET.

Abstract

Positron emission tomography (PET) is a well-established functional imaging technique for diagnosing brain disorders. However, PET's high costs and radiation exposure limit its widespread use. In contrast, magnetic resonance imaging (MRI) does not have these limitations. Although it also captures neurodegenerative changes, MRI is a less sensitive diagnostic tool than PET. To close this gap, we aim to generate synthetic PET from MRI. Herewith, we introduce PASTA, a novel pathology-aware image translation framework based on conditional diffusion models. Compared to the state-of-the-art methods, PASTA excels in preserving both structural and pathological details in the target modality, which is achieved through its highly interactive dual-arm architecture and multi-modal condition integration. A cycle exchange consistency and volumetric generation strategy elevate PASTA's capability to produce high-quality 3D PET scans. Our qualitative and quantitative results confirm that the synthesized PET scans from PASTA not only reach the best quantitative scores but also preserve the pathology correctly. For Alzheimer's classification, the performance of synthesized scans improves over MRI by 4%, almost reaching the performance of actual PET. Code is available at https://github.com/ai-med/PASTA.
Paper Structure (21 sections, 6 equations, 6 figures, 6 tables)

This paper contains 21 sections, 6 equations, 6 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Proposed Method
  4. Conditioner Arm:
  5. Adaptive Conditional Module:
  6. Denoiser Arm
  7. Cycle Exchange Consistency (CycleEx):
  8. Volumetric Generation:
  9. Experiments
  10. Results
  11. Qualitative Results and Clinical Assessment
  12. Quantitative Results
  13. Classification Results for AD Diagnosis
  14. Ablation Study
  15. Conclusion
  16. ...and 6 more sections

Figures (6)

  • Figure 1: For Alzheimer's disease, PET shows reduced glucose uptake in the temporoparietal lobe (bottom circles), mirroring atrophy on MRI with higher sensitivity. While state-of-the-art diffusion models fail to recover such pathology in the synthesized PET, PASTA achieves significant improvements.
  • Figure 2: Overall structure of PASTA.
  • Figure 3: Cycle exchange consistency (CycleEx) strategy of PASTA.
  • Figure 4: Qualitative results for a normal subject (top) and an AD patient (middle) with magnified specific pathology (hypometabolism in temporoparietal lobes).
  • Figure A.1: Data preprocessing steps for MRI and PET. Both scans have been preprocessed from the Alzheimer’s disease neuroimaging initiative (ADNI).
  • ...and 1 more figures