Related Experiment Video
Updated: Jul 27, 2025
10:02
Whole-body PET/MRI of Pediatric Patients: The Details That Matter
Published on: December 19, 2017
14.6K
An automatic pipeline for PET/MRI attenuation correction validation in the brain.
Mahdjoub Hamdi1, Chunwei Ying2, Hongyu An2
1Washington University In St Louis: Washington University in St Louis.
Research Square
|June 9, 2023
Summary
This study introduces an automated pipeline to assess MRI-based attenuation correction (MRAC) for PET/MRI accuracy. The deep learning approach (DL-DIXON AC) demonstrated the lowest quantitative bias, outperforming other MRAC methods.
Area of Science:
- Neuroimaging
- Medical Physics
Background:
- Accurate PET/MRI quantitative analysis in neurology is hindered by PET attenuation correction (AC) accuracy.
- MRI-based attenuation correction (MRAC) offers a promising alternative to CT-based AC (CTAC).
Approach:
- An automated pipeline was developed using a synthetic lesion insertion tool and FreeSurfer for ROI generation.
- Four MRAC techniques (DIXON, DIXONbone, UTE, DL-DIXON) were evaluated against CTAC using patient PET/MRI data.
- Quantitative accuracy was assessed by comparing MRAC to CTAC bias in simulated lesions and brain ROIs, with and without background activity.
Key Points:
- The pipeline accurately and consistently evaluated MRAC methods for synthetic lesions and ROIs.
- DL-DIXON AC exhibited the lowest MRAC to CTAC bias across all tested scenarios.
- DIXON AC showed the highest bias, followed by UTE and DIXONbone.
Conclusions:
- The developed pipeline enables robust evaluation of novel MRAC techniques without requiring PET emission data.
- DL-DIXON AC presents a highly accurate MRAC approach for quantitative PET/MRI in neurological applications.
