Probabilistic Air Segmentation and Sparse Regression Estimated Pseudo CT for PET/MR Attenuation Correction
Yasheng Chen1, Meher Juttukonda, Yi Su
1From the Biomedical Research Imaging Center (Y.C., Y.Z.L., W.L., D.S., D.L., H.A.), Department of Radiology (Y.C., Y.Z.L., W.L., D.S., H.A.), and Department of Biomedical Engineering (M.J., Y.Z.L., W.L., D.L., H.A.), University of North Carolina at Chapel Hill, 106 Mason Farm Rd, CB 7513, Chapel Hill, NC 27599; and Mallinckrodt Institute of Radiology, Washington University School of Medicine, St Louis, Mo (Y.S., T.B., B.G.R.).
A new probabilistic air segmentation and sparse regression (PASSR) method significantly improves positron emission tomography (PET) attenuation correction accuracy in brain PET/MR imaging by generating accurate pseudo-CT images.
Area of Science:
- Medical Imaging
- Radiology
- Nuclear Medicine
Background:
- Accurate attenuation correction is crucial for quantitative positron emission tomography (PET) imaging.
- Integrating PET with magnetic resonance (MR) imaging offers complementary anatomical and functional information.
- Traditional CT-based attenuation correction in PET/MR is challenging due to hardware limitations.
Purpose of the Study:
- To develop and validate a novel pseudo-CT generation method for PET attenuation correction in brain PET/MR imaging.
- To estimate pseudo-CT images from T1-weighted MR and atlas CT data.
- To reduce PET image errors caused by attenuation correction inaccuracies.
Main Methods:
- Developed a probabilistic air segmentation and sparse regression (PASSR) method for pseudo-CT estimation.
- Utilized T1-weighted MR images and atlas CT data for regression-based CT number estimation in non-air regions.
- Acquired PET/MR/CT data from 20 subjects for method validation.
- Compared PASSR with Dixon segmentation, CT segmentation, and mean atlas methods using Mean Absolute Percentage Error (MAPE) on PET images.
Main Results:
- The PASSR method achieved significantly lower MAPE in whole brain (2.42% ± 1.0), gray matter (3.28% ± 0.93), and white matter (2.16% ± 1.75) compared to other methods (P < .01).
- PASSR demonstrated superior accuracy, with 68.0% ± 16.5, 85.8% ± 12.9, and 96.0% ± 2.5 of whole-brain volume within ±2%, ±5%, and ±10% error, respectively.
- These error margins were significantly better than those achieved by Dixon, CT segmentation, and mean atlas methods (P < .01).
Conclusions:
- The PASSR method significantly outperforms existing methods for PET attenuation correction in brain PET/MR imaging.
- PASSR effectively reduces PET image errors by providing more accurate attenuation maps.
- This novel approach enhances the quantitative accuracy of PET imaging in hybrid PET/MR systems.
More Related Videos
06:53Management of Respiratory Motion Artefacts in 18F-fluorodeoxyglucose Positron Emission Tomography using an Amplitude-Based Optimal Respiratory Gating Algorithm
Published on: July 23, 2020
14:08Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013