Tehran University of Medical Sciences

Science Communicator Platform

Share By
Comparative Assessment of U-Net and Pix2pix for Applying Direct Attenuation Correction in the Image Domain in 68 Ga-Psma Pet/Ct Imaging Publisher



Hamidiyan N ; Ahangari HT ; Ghafarian P ; Arabi H ; Ay MR
Authors

Source: 32nd National and 10th International Iranian Conference on Biomedical Engineering, ICBME 2025 Published:2025


Abstract

Accurate quantitative positron emission tomography (PET) imaging needs effective attenuation correction (AC). This remains a challenge in dedicated PET systems that lack concurrent computed tomography (CT). Recent research has investigated deep learning (DL) methods for AC, but direct comparisons between models are still limited. This study systematically compares the performance of two widely used DL architectures, U-Net and Pix2Pix, for direct AC of whole-body 68 Ga-PSMA PET images using a consistent set of 95 patient data sets. Both models are evaluated under identical conditions. For each data set, CT-based attenuation-corrected PET (PET-CTAC) was used as the reference. Quantitative evaluation included mean error (ME) of mean of standardized uptake value (SUVmean), normalized root mean square error (NRMSE), structural similarity index (SSIM), and peak signal-to-noise ratio (PSNR). Both U-Net and Pix2Pix generate PET images with visual quality similar to PET-CTAC, but Pix2Pix generally showed better quantitative metrics. Specifically, UNet achieved ME, NRMSE, SSIM, and PSNR values of 0.037 ± 0.02,0.006 ± 0.005,12.88 ± 2.73, and 0.98 ± 0.14, respectively, whereas Pix2Pix achieved 0.015 ± 0.015,0.005 ± 0.004,13.93 ± 2.48, and 0.99 ± 0.004. Statistical analyses are conducted using paired t-test or Wilcoxon signed-rank tests, selected based on the normality of the data, demonstrated that Pix2Pix produced SUV estimates closer to those of PET-CTAC, with lower bias and variability than U-Net. In conclusion, both DL models enabled direct AC of whole-body 68 Ga-PSMA PET, but Pix2Pix provided more accurate and reliable AC when the two models were directly compared, indicating Pix2Pix is the stronger candidate for clinical use in dedicated PET systems without CT imaging. © 2025 IEEE.
Related Docs
View other Related Docs
1. Deep-Jasc: Joint Attenuation and Scatter Correction in Whole-Body 18F-Fdg Pet Using a Deep Residual Network, European Journal of Nuclear Medicine and Molecular Imaging (2020)
2. Deep-Learning-Based Attenuation Correction For68 Ga-Dotatate Whole-Body Pet Imaging: A Dual-Center Clinical Study; [68Ga-Dotatate Tum Vucut Pet Goruntuleme Icin Derin Ogrenme Tabanli Atenuasyon Duzeltmesi: Cift Merkezli Bir Klinik Calisma], Molecular Imaging and Radionuclide Therapy (2024)
3. Deep Active Learning Model for Adaptive Pet Attenuation and Scatter Correction in Multi-Centric Studies, 2021 IEEE Nuclear Science Symposium and Medical Imaging Conference Record# NSS/MIC 2021 and 28th International Symposium on Room-Temperature Semiconductor Detectors# RTSD 2022 (2021)
Experts (# of related papers)
View all Related Experts
Parham Geramifar (7)
Peyman Sheikhzadeh (5)
Other Related Docs
4. Deep Adaptive Transfer Learning for Site-Specific Pet Attenuation and Scatter Correction From Multi-National/Institutional Datasets, 2022 IEEE NSS/MIC RTSD - IEEE Nuclear Science Symposium# Medical Imaging Conference and Room Temperature Semiconductor Detector Conference (2022)
5. Prediction of the Gleason Score of Prostate Cancer Patients Using 68Ga-Psma-Pet/Ct Radiomic Models, Journal of Medical and Biological Engineering (2024)
6. Direct Attenuation Correction of Brain Pet Images Using Only Emission Data Via a Deep Convolutional Encoder-Decoder (Deep-Dac), European Radiology (2019)
7. A Ct-Free Deep-Learning-Based Attenuation and Scatter Correction for Copper-64 Pet in Different Time-Point Scans, Radiological Physics and Technology (2025)
8. Enhanced Direct Joint Attenuation and Scatter Correction of Whole-Body Pet Images Via Context-Aware Deep Networks, Zeitschrift fur Medizinische Physik (2024)
9. Generation of Synthetic Ct From Mri for Mri-Based Attenuation Correction of Brain Pet Images Using Radiomics and Machine Learning, Medical Physics (2025)
10. Standardization and Optimization of Siemens Biograph Truepoint Pet/Ct Acquisition and Reconstruction Parameters: Simultaneous Qualitative and Quantitative Assessments, Iranian Journal of Nuclear Medicine (2024)
11. A Novel Attention-Based Convolutional Neural Network for Joint Denoising and Partial Volume Correction of Low-Dose Pet Images, 2021 IEEE Nuclear Science Symposium and Medical Imaging Conference Record# NSS/MIC 2021 and 28th International Symposium on Room-Temperature Semiconductor Detectors# RTSD 2022 (2021)