Nghiên cứu hiệu quả mô hình hiệu chỉnh nhiễu suy giảm trong SPECT xạ hình tưới máu cơ tim

Abstract

Objective: To evaluate the effectiveness of using deep learning model with attenuation correction in SPECT myocardial perfusion imaging. Subject and method: Use the 3DUnet-GAN network to attenuation correction for SPECT without attenuation correction (NC) images to generate attenuation corrected (AC) SPECT images. The image generated from the AI model is compared with the SPECT/CT image (TrueAC). Compare and evaluate the results of the model used with ResNet model and Chang AC method, the coefficient of uniform decline in body contour. Evaluate model-generated images with clinical effectiveness. Result: Overall, the deep learning solution exhibited good agreement with the CT-based AC, noticeably outperforming the Chang method. The ResNet and 3DUNet-GAN models resulted in the ME (count) of -6.99 ± 16.72 and -4.41 ± 11.8 and SSIM of 0.99 ± 0.04 and 0.98 ± 0.05, respectively. While the Chang approach led to ME and SSIM of 25.52 ± 33.98 and 0.93 ± 0.09, respectively. Similarly, the clinical evaluation revealed a mean TPD of 12.21 ± 8.28 and 12.00 ± 9.21 for the ResNet and UNet models, respectively, compared to 11.95 ± 9.37 obtained from the reference SPECT CT-AC images. On the other hand, the Chang approach led to a mean TPD of 14.26 ± 8.19. Conclusion: The method of using computer deep learning to attenuation correction SPECT images potential to be used for facilities that only use SPECT machines, assisting doctors in diagnosing images and reducing the pressure on work intensity for medical staff.