Theruvath Ashok J, Siedek Florian, Yerneni Ketan, Muehe Anne M, Spunt Sheri L, Pribnow Allison, Moseley Michael, Lu Ying, Zhao Qian, Gulaka Praveen, Chaudhari Akshay, Daldrup-Link Heike E
Department of Radiology, Molecular Imaging Program at Stanford (A.J.T., F.S., K.Y., A.M.M., M.M., A.C., H.E.D.L.), Department of Pediatrics, Division of Hematology/Oncology, Lucile Packard Children's Hospital (S.L.S., A.P., H.E.D.L.), and Department of Biomedical Data Science (Y.L., Q.Z.), Stanford University, 725 Welch Rd, Stanford, CA 94304; and Subtle Medical, Menlo Park, Calif (P.G.).
Radiol Artif Intell. 2021 Oct 6;3(6):e200232. doi: 10.1148/ryai.2021200232. eCollection 2021 Nov.
To investigate if a deep learning convolutional neural network (CNN) could enable low-dose fluorine 18 (F) fluorodeoxyglucose (FDG) PET/MRI for correct treatment response assessment of children and young adults with lymphoma.
In this secondary analysis of prospectively collected data (ClinicalTrials.gov identifier: NCT01542879), 20 patients with lymphoma (mean age, 16.4 years ± 6.4 [standard deviation]) underwent F-FDG PET/MRI between July 2015 and August 2019 at baseline and after induction chemotherapy. Full-dose F-FDG PET data (3 MBq/kg) were simulated to lower F-FDG doses based on the percentage of coincidence events (representing simulated 75%, 50%, 25%, 12.5%, and 6.25% F-FDG dose [hereafter referred to as 75%, 50%, 25%, 12.5%, and 6.25%, respectively]). A U.S. Food and Drug Administration-approved CNN was used to augment input simulated low-dose scans to full-dose scans. For each follow-up scan after induction chemotherapy, the standardized uptake value (SUV) response score was calculated as the maximum SUV (SUV) of the tumor normalized to the mean liver SUV; tumor response was classified as adequate or inadequate. Sensitivity and specificity in the detection of correct response status were computed using full-dose PET as the reference standard.
With decreasing simulated radiotracer doses, tumor SUV increased. A dose below 75% of the full dose led to erroneous upstaging of adequate responders to inadequate responders (43% [six of 14 patients] for 75%; 93% [13 of 14 patients] for 50%; and 100% [14 of 14 patients] below 50%; < .05 for all). CNN-enhanced low-dose PET/MRI scans at 75% and 50% enabled correct response assessments for all patients. Use of the CNN augmentation for assessing adequate and inadequate responses resulted in identical sensitivities (100%) and specificities (100%) between the assessment of 100% full-dose PET, augmented 75%, and augmented 50% images.
CNN enhancement of PET/MRI scans may enable 50% F-FDG dose reduction with correct treatment response assessment of children and young adults with lymphoma. Pediatrics, PET/MRI, Computer Applications Detection/Diagnosis, Lymphoma, Tumor Response, Whole-Body Imaging, Technology AssessmentClinical trial registration no: NCT01542879 © RSNA, 2021.
研究深度学习卷积神经网络（CNN）是否能够使低剂量氟18（F）氟脱氧葡萄糖（FDG）PET/MRI用于正确评估淋巴瘤儿童和青年患者的治疗反应。
在对前瞻性收集的数据进行的这项二次分析中（ClinicalTrials.gov标识符：NCT01542879），20例淋巴瘤患者（平均年龄16.4岁±6.4[标准差]）于2015年7月至2019年8月在基线期及诱导化疗后接受了F-FDG PET/MRI检查。基于符合事件的百分比（分别代表模拟的75%、50%、25%、12.5%和6.25%的F-FDG剂量[以下分别简称为75%、50%、25%、12.5%和6.25%]）模拟全剂量F-FDG PET数据（3MBq/kg）以降低F-FDG剂量。使用美国食品药品监督管理局批准的CNN将输入的模拟低剂量扫描增强为全剂量扫描。对于诱导化疗后的每次随访扫描，标准化摄取值（SUV）反应评分计算为肿瘤的最大SUV（SUVmax）除以肝脏平均SUV；肿瘤反应分为充分或不充分。以全剂量PET作为参考标准计算检测正确反应状态的敏感性和特异性。
随着模拟放射性示踪剂剂量的降低，肿瘤SUV增加。全剂量的75%以下的剂量导致将反应充分的患者错误地分期为反应不充分的患者（75%时为43%[14例患者中的6例]；50%时为93%[14例患者中的13例]；50%以下时为100%[14例患者中的14例]；所有情况P<0.05）。75%和50%剂量的CNN增强低剂量PET/MRI扫描能够对所有患者进行正确的反应评估。使用CNN增强来评估反应充分和不充分的情况，在100%全剂量PET、增强75%和增强50%图像的评估之间，敏感性（100%）和特异性（100%）相同。
PET/MRI扫描的CNN增强可能使F-FDG剂量降低50%，同时对淋巴瘤儿童和青年患者进行正确的治疗反应评估。儿科学、PET/MRI、计算机应用检测/诊断、淋巴瘤、肿瘤反应、全身成像、技术评估临床试验注册号：NCT01542879 ©RSNA，2021。
