Automated quantification of brain PET in PET/CT using deep learning-based CT-to-MR translation: a feasibility study.
Authors
Affiliations (5)
Affiliations (5)
- Department of Artificial Intelligence, Yonsei University, Seoul, Republic of Korea.
- Department of Computer Science, Yonsei University, Seoul, Republic of Korea.
- Department of Nuclear Medicine, Yonsei University College of Medicine, Seoul, Republic of Korea.
- Department of Nuclear Medicine, Yonsei University College of Medicine, Seoul, Republic of Korea. [email protected].
- Department of Radiology, University of Texas Southwestern, Dallas, TX, USA.
Abstract
Quantitative analysis of PET images in brain PET/CT relies on MRI-derived regions of interest (ROIs). However, the pairs of PET/CT and MR images are not always available, and their alignment is challenging if their acquisition times differ considerably. To address these problems, this study proposes a deep learning framework for translating CT of PET/CT to synthetic MR images (MR<sub>SYN</sub>) and performing automated quantitative regional analysis using MR<sub>SYN</sub>-derived segmentation. In this retrospective study, 139 subjects who underwent brain [<sup>18</sup>F]FBB PET/CT and T1-weighted MRI were included. A U-Net-like model was trained to translate CT images to MR<sub>SYN</sub>; subsequently, a separate model was trained to segment MR<sub>SYN</sub> into 95 regions. Regional and composite standardised uptake value ratio (SUVr) was calculated in [<sup>18</sup>F]FBB PET images using the acquired ROIs. For evaluation of MR<sub>SYN</sub>, quantitative measurements including structural similarity index measure (SSIM) were employed, while for MR<sub>SYN</sub>-based segmentation evaluation, Dice similarity coefficient (DSC) was calculated. Wilcoxon signed-rank test was performed for SUVrs computed using MR<sub>SYN</sub> and ground-truth MR (MR<sub>GT</sub>). Compared to MR<sub>GT</sub>, the mean SSIM of MR<sub>SYN</sub> was 0.974 ± 0.005. The MR<sub>SYN</sub>-based segmentation achieved a mean DSC of 0.733 across 95 regions. No statistical significance (P > 0.05) was found for SUVr between the ROIs from MR<sub>SYN</sub> and those from MR<sub>GT</sub>, excluding the precuneus. We demonstrated a deep learning framework for automated regional brain analysis in PET/CT with MR<sub>SYN</sub>. Our proposed framework can benefit patients who have difficulties in performing an MRI scan.