Related Experiment Video
Updated: Jun 12, 2026
10:02
Whole-body PET/MRI of Pediatric Patients: The Details That Matter
Published on: December 19, 2017
MR-based attenuation correction for hybrid PET-MR brain imaging systems using deformable image registration.
Eduard Schreibmann1, Jonathon A Nye, David M Schuster
1Department of Radiation Oncology, School of Medicine, Emory University, Atlanta, Georgia 30322, USA. edi@radonc.emory.org
Medical Physics
|June 10, 2010
Summary
Accurate attenuation correction for PET-MR scanners is now possible using a novel deformable model that personalizes an atlas CT to patient MR data. This method eliminates the need for additional CT scans, ensuring clinical accuracy for hybrid imaging.
Area of Science:
- Medical Imaging
- Radiology
- Biomedical Engineering
Background:
- Combined positron emission tomography (PET) and magnetic resonance (MR) scanners offer simultaneous molecular and anatomical imaging, revolutionizing clinical practice.
- PET-MR is becoming the preferred modality for neurological studies, cancer imaging, stroke assessment, and stem cell research.
- A key challenge for PET-MR implementation is the lack of direct attenuation correction maps due to space and cost limitations.
Purpose of the Study:
- To present a method for generating accurate, patient-specific PET attenuation coefficient maps for head imaging using combined PET-MR scanners.
- To address the challenge of obtaining attenuation correction data without requiring a separate computed tomography (CT) scan.
Main Methods:
- A multimodality optical flow deformable model was developed to establish voxel-to-voxel correspondence between a CT atlas and patient MR images.
- The CT atlas was warped using the derived deformation field to create a simulated CT image matching patient anatomy.
- This simulated CT image serves as a basis for attenuation correction in PET-MR imaging.
Main Results:
- The deformable registration model achieved accurate mapping from an atlas CT to individual patient MR datasets in 17 brain tumor cases.
- Geometrical accuracy was confirmed, with mean distances of 1.26 mm (external contour) and 2.15 mm (bony anatomy) between simulated and true CT.
- Voxel-to-voxel comparison of Hounsfield units (HUs) showed a mean difference of less than 2 HU, indicating high accuracy.
Conclusions:
- Patient-specific attenuation correction for hybrid PET-MR scanners can be efficiently achieved by individualizing an atlas CT to MR data via a deformable model.
- This automated method provides clinical accuracy without the need for an additional CT scan.
- The approach simplifies PET-MR workflow and enhances its clinical applicability.
