Department of Medical Physics, School of Medicine, University of Patras, Rion, Greece.
School of Electrical and Computer Engineering, University of Patras, Rion, Greece.
Med Phys. 2021 Nov;48(11):7427-7438. doi: 10.1002/mp.15270. Epub 2021 Oct 21.
Radioembolization with Y microspheres is a treatment approach for liver cancer. Currently, employed dosimetric calculations exhibit low accuracy, lacking consideration of individual patient, and tissue characteristics.
The purpose of the present study was to employ deep learning (DL) algorithms to differentiate patterns of pretreatment distribution of Tc-macroaggregated albumin on SPECT/CT and post-treatment distribution of Y microspheres on PET/CT and to accurately predict how the Y-microspheres will be distributed in the liver tissue by radioembolization therapy.
Data for 19 patients with liver cancer (10 with hepatocellular carcinoma, 5 with intrahepatic cholangiocarcinoma, 4 with liver metastases) who underwent radioembolization with Y microspheres were used for the DL training. We developed a 3D voxel-based variation of the Pix2Pix model, which is a special type of conditional GANs designed to perform image-to-image translation. SPECT and CT scans along with the clinical target volume for each patient were used as inputs, as were their corresponding post-treatment PET scans. The real and predicted absorbed PET doses for the tumor and the whole liver area were compared. Our model was evaluated using the leave-one-out method, and the dose calculations were measured using a tissue-specific dose voxel kernel.
The comparison of the real and predicted PET/CT scans showed an average absorbed dose difference of 5.42% ± 19.31% and 0.44% ± 1.64% for the tumor and the liver area, respectively. The average absorbed dose differences were 7.98 ± 31.39 Gy and 0.03 ± 0.25 Gy for the tumor and the non-tumor liver parenchyma, respectively. Our model had a general tendency to underpredict the dosimetric results; the largest differences were noticed in one case, where the model underestimated the dose to the tumor area by 56.75% or 72.82 Gy.
The proposed deep-learning-based pretreatment planning method for liver radioembolization accurately predicted Y microsphere biodistribution. Its combination with a rapid and accurate 3D dosimetry method will render it clinically suitable and could improve patient-specific pretreatment planning.
放射性栓塞治疗肝癌采用 Y 微球。目前，所采用的剂量计算方法准确性较低，未考虑患者个体和组织特征。
本研究旨在采用深度学习（DL）算法区分 SPECT/CT 预处理时 Tc-巨聚合白蛋白分布模式和 PET/CT 后 Y 微球分布模式，并准确预测 Y 微球在放射性栓塞治疗中在肝组织内的分布情况。
使用 19 例接受 Y 微球放射性栓塞治疗的肝癌患者（肝细胞癌 10 例，肝内胆管细胞癌 5 例，肝转移癌 4 例）的数据进行 DL 训练。我们开发了一种基于 3D 体素的 Pix2Pix 模型变体，这是一种特殊类型的条件生成对抗网络，用于执行图像到图像的转换。将每位患者的 SPECT 和 CT 扫描以及临床靶区作为输入，同时将其相应的治疗后 PET 扫描作为输入。比较肿瘤和整个肝区的真实和预测吸收性 PET 剂量。采用留一法评估模型，使用组织特异性剂量体素核评估剂量计算。
真实和预测的 PET/CT 扫描比较显示，肿瘤和肝脏区域的平均吸收剂量差异分别为 5.42%±19.31%和 0.44%±1.64%。肿瘤和非肿瘤性肝实质的平均吸收剂量差异分别为 7.98±31.39 Gy 和 0.03±0.25 Gy。模型普遍存在低估剂量学结果的趋势；在一个病例中，模型低估了肿瘤区域的剂量 56.75%或 72.82 Gy，差异最大。
提出的用于肝脏放射性栓塞的基于深度学习的预处理计划方法能够准确预测 Y 微球的生物分布。结合快速、准确的 3D 剂量计算方法，将使其在临床上适用，并可改善个体化的预处理计划。
