From the Department of Product Development-Personalized HealthCare Imaging (A.P.K., Z.S., T.B., R.A.D.C.), Clinical Imaging Group, gRED (D.C., A.d.C.), and DevSci OMNI-Biomarker Development (X.J.), Genentech, 600 E Grand Ave, South San Francisco, CA 94080; and Global Product Development Medical Affairs, Neuroscience, F. Hoffmann-La Roche, Basel, Switzerland (L.G.).
Radiology. 2022 Mar;302(3):662-673. doi: 10.1148/radiol.211528. Epub 2021 Dec 14.
Background Deep learning-based segmentation could facilitate rapid and reproducible T1 lesion load assessments, which is crucial for disease management in multiple sclerosis (MS). T1 unenhancing and contrast-enhancing lesions in MS are those that enhance or do not enhance after administration of a gadolinium-based contrast agent at T1-weighted MRI. Purpose To develop deep learning models for automated assessment of T1 unenhancing and contrast-enhancing lesions; to investigate if joint training improved performance; to reproduce a known ocrelizumab treatment response; and to evaluate the association of baseline T1-weighted imaging metrics with clinical outcomes in relapsing MS clinical trials. Materials and Methods Joint and individual deep learning models (U-Nets) were developed retrospectively on multimodal MRI data sets from large multicenter OPERA trials of relapsing MS (August 2011 to May 2015). The joint model included cross-network connections and a combined loss function. Models were trained on OPERA I data sets with three-fold cross-validation. OPERA II data sets were the internal test set. Dice coefficients, lesion true-positive and false-positive rates, and areas under the receiver operating characteristic curve (AUCs) were used to evaluate model performance. Association of baseline imaging metrics with clinical outcomes was assessed with Cox proportional hazards models. Results A total of 796 patients (3030 visits; mean age, 37 years ± 9; 521 women) from the OPERA II trial were evaluated. The joint model achieved a mean Dice coefficient of 0.77 and 0.74, lesion true-positive rate of 0.88 and 0.86, and lesion false-positive rate of 0.04 and 0.19 for T1 contrast-enhancing and T1 unenhancing lesion segmentation, respectively. Joint training improved performance for smaller T1 contrast-enhancing lesions (≤0.06 mL; individual training AUC: 0.75; joint training AUC: 0.87; < .001). A significant ocrelizumab treatment effect ( < .001) was seen in reducing the mean number of T1 contrast-enhancing lesions at 24, 48, and 96 weeks (manual assessment at 24 weeks: 10 lesions in 366 patients with ocrelizumab, 141 lesions in 355 patients with interferon, 93% reduction; manual assessment at 48 weeks: six lesions in 355 patients with ocrelizumab, 150 lesions in 317 patients with interferon, 96% reduction; manual assessment at 96 weeks: five lesions in 340 patients with ocrelizumab, 157 lesions in 294 patients with interferon, 97% reduction; joint model assessment at 24 weeks: 19 lesions in 365 patients with ocrelizumab, 128 lesions in 354 patients with interferon, 86% reduction; joint model assessment at 48 weeks: 14 lesions in 355 patients with ocrelizumab, 121 lesions in 317 patients with interferon, 90% reduction; joint model assessment at 96 weeks: 10 lesions in 340 patients with ocrelizumab, 144 lesions in 294 patients with interferon, 94% reduction) and the mean number of new T1 unenhancing lesions across all follow-up examinations (manual assessment: 504 lesions in 1060 visits for ocrelizumab-treated patients, 1438 lesions in 965 visits for interferon-treated patients, 68% reduction; joint model assessment: 205 lesions in 1053 visits for ocrelizumab-treated patients, 661 lesions in 957 visits for interferon-treated patients, 78% reduction). Baseline T1 unenhancing total lesion volume was associated with clinical outcomes (manual hazard ratio [HR]: 1.12, = .02; joint model HR: 1.11, = .03). Conclusion Joint architecture and training improved segmentation of MRI T1 contrast-enhancing multiple sclerosis lesions, and both deep learning models had sufficiently high performance to detect an ocrelizumab treatment response consistent with manual assessments. ClinicalTrials.gov: NCT01247324 and NCT01412333 © RSNA, 2021 See also the editorial by Talbott in this issue.
基于深度学习的分割技术可以促进快速、可重复的 T1 病变负荷评估，这对于多发性硬化症（MS）的疾病管理至关重要。MS 的 T1 无增强和增强病变是指在 T1 加权 MRI 中使用钆基对比剂后增强或不增强的病变。目的：开发用于自动评估 T1 无增强和增强病变的深度学习模型；研究联合训练是否可以提高性能；再现已知的奥瑞珠单抗治疗反应；以及评估复发型 MS 临床试验中基线 T1 加权成像指标与临床结局的相关性。材料与方法：回顾性地在大型多中心 OPERA 试验的复发型 MS 多模态 MRI 数据集上开发了联合和单独的深度学习模型（U-Nets）（2011 年 8 月至 2015 年 5 月）。联合模型包括跨网络连接和联合损失函数。使用三折交叉验证在 OPERA I 数据集上对模型进行训练。OPERA II 数据集是内部测试集。使用 Dice 系数、病变真阳性和假阳性率以及受试者工作特征曲线（AUC）下面积来评估模型性能。使用 Cox 比例风险模型评估基线成像指标与临床结局的相关性。结果：在 OPERA II 试验中，共评估了 796 例患者（3030 次就诊；平均年龄 37 岁±9；521 例女性）。联合模型的平均 Dice 系数分别为 0.77 和 0.74，病变真阳性率分别为 0.88 和 0.86，病变假阳性率分别为 0.04 和 0.19，用于 T1 对比增强和 T1 无增强病变分割。较小的 T1 对比增强病变（≤0.06 mL；个体训练 AUC：0.75；联合训练 AUC：0.87；<.001）的联合训练提高了性能。在 24、48 和 96 周时，奥瑞珠单抗治疗显著降低了 T1 对比增强病变的平均数量（24 周时手动评估：奥瑞珠单抗治疗组 366 例患者中有 10 个病变，干扰素治疗组 355 例患者中有 141 个病变，减少 93%；48 周时手动评估：奥瑞珠单抗治疗组 355 例患者中有 6 个病变，干扰素治疗组 317 例患者中有 150 个病变，减少 96%；96 周时手动评估：奥瑞珠单抗治疗组 340 例患者中有 5 个病变，干扰素治疗组 294 例患者中有 157 个病变，减少 97%；联合模型评估 24 周时：奥瑞珠单抗治疗组 365 例患者中有 19 个病变，干扰素治疗组 354 例患者中有 128 个病变，减少 86%；联合模型评估 48 周时：奥瑞珠单抗治疗组 355 例患者中有 14 个病变，干扰素治疗组 317 例患者中有 121 个病变，减少 90%；联合模型评估 96 周时：奥瑞珠单抗治疗组 340 例患者中有 10 个病变，干扰素治疗组 294 例患者中有 144 个病变，减少 94%）和所有随访检查的新 T1 无增强病变的平均数量（手动评估：奥瑞珠单抗治疗组 1060 次就诊中有 504 个病变，干扰素治疗组 965 次就诊中有 1438 个病变，减少 68%；联合模型评估：奥瑞珠单抗治疗组 1053 次就诊中有 205 个病变，干扰素治疗组 957 次就诊中有 661 个病变，减少 78%）。基线 T1 无增强总病变体积与临床结局相关（手动风险比[HR]：1.12， =.02；联合模型 HR：1.11， =.03）。结论：联合架构和训练提高了 MRI T1 对比增强多发性硬化病变的分割性能，并且两种深度学习模型的性能都足够高，可以检测到与手动评估一致的奥瑞珠单抗治疗反应。临床试验注册：NCT01247324 和 NCT01412333 © RSNA，2021 还请参阅本期的 Talbott 社论。
