AUTHOR=Du Yu , Jiang Han , Lin Ching-Ni , Peng Zhengyu , Sun Jingzhang , Chiu Pai-Yi , Hung Guang-Uei , Mok Greta S. P. TITLE=Generative adversarial network-based attenuation correction for 99mTc-TRODAT-1 brain SPECT JOURNAL=Frontiers in Medicine VOLUME=Volume 10 - 2023 YEAR=2023 URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2023.1171118 DOI=10.3389/fmed.2023.1171118 ISSN=2296-858X ABSTRACT=Background: Attenuation correction (AC) is an important correction method to improve the quantification accuracy of dopamine transporter (DAT) SPECT. Chang’s method was developed for AC (Chang-AC) when CT-based AC was not available, assuming uniform attenuation coefficients inside body contour. This study aims to evaluate Chang-AC and different deep learning (DL)-based AC approaches on 99mTc-TRODAT-1 brain SPECT using clinical patient data on two different scanners. Methods: Two hundred and sixty patients underwent 99mTc-TRODAT-1 SPECT/CT scans from two different scanners (scanner A and scanner B) were retrospectively recruited. One hundred and twenty projections were reconstructed by OS-EM method with dual-energy scatter correction, with or without CT-AC. We implemented a 3D conditional generative adversarial network (cGAN) for the indirect (DL-ACµ) and direct (DL-AC) DL-based AC methods, estimating attenuation maps and attenuation corrected SPECT images from non-attenuation corrected (NAC) SPECT respectively. We further applied cross-scanner training (cDL-ACµ and cDL-AC) and merged the datasets from two scanners for ensemble training (eDL-ACµ and eDL-AC). The estimated attenuation maps from (c/e)DL-ACµ were then used in reconstruction for AC purpose. Chang’s method was also implemented for comparison. Normalized mean square error (NMSE), structural similarity index (SSIM), specific uptake ratio (SUR) and asymmetric index (%ASI) of the striatum were calculated for different AC methods. Results: The NMSE for Chang’s method, DL-ACµ, DL-AC, cDL-ACµ, cDL-AC, eDL-ACµ and eDL-AC are 0.0406±0.0445, 0.0059±0.0035, 0.0099±0.0066, 0.0253±0.0102, 0.0369±0.0124, 0.0098±0.0035 and 0.0162±0.0118 for scanner A, and 0.0579±0.0146, 0.0055±0.0034, 0.0063±0.0028, 0.0235±0.0085, 0.0349±0.0086, 0.0115±0.0062 and 0.0117±0.0038 for scanner B. The SUR and %ASI results for DL-ACµ are closest to CT-AC, followed by DL-AC, eDL-ACµ, cDL-ACµ, cDL-AC, eDL-AC, Chang’s method and NAC. Conclusions: All DL-based AC methods are superior to Chang-AC. DL-ACµ is superior to DL-AC. Scanner specific training is superior to cross-scanner and ensemble training. DL-based AC methods are feasible and robust for 99mTc-TRODAT-1 brain SPECT. Keywords: deep learning, generative adversarial network, attenuation correction, dopamine transporter SPECT, 99mTc-TRODAT-1