Department of Dermatology, The Third Affiliated Hospital of Sun Yat-Sen University, Guangzhou, Guangdong, China; Department of Dermatology, Cosmetology and Venereology, Shenzhen Hospital, Southern Medical University, Shenzhen, Guangdong, China.
Department of Nuclear Medicine, The Seventh Affiliated Hospital, Sun Yat-Sen University, Shenzhen, China; Department of Radiology, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, Guangzhou, Guangdong, China.
Radiother Oncol. 2023 Sep;186:109737. doi: 10.1016/j.radonc.2023.109737. Epub 2023 Jun 12.
Dermatofibrosarcoma protuberans (DFSP) is characterized by locally invasive growth patterns and high local recurrence rates. Accurately identifying patients with high local recurrence risk may benefit patients during follow-up and has potential value for making treatment decisions. This study aimed to investigate whether machine learning-based radiomics models could accurately predict the local recurrence of primary DFSP after surgical treatment.
This retrospective study included a total of 146 patients with DFSP who underwent MRI scans between 2010 and 2016 from two different institutions: institution 1 (n = 104) for the training set and institution 2 (n = 42) for the external test set. Three radiomics random survival forest (RSF) models were developed using MRI images. Additionally, the performance of the Ki67 index was compared with the three RSF models in the external validation set.
The average concordance index (C-index) scores of the RSF models based on fat-saturation T2W (FS-T2W) images, fat-saturation T1W with gadolinium contrast (FS-T1W + C) images, and both FS-T2W and FS-T1W + C images from 10-fold cross-validation in the training set were 0.855 (95% CI: 0.629, 1.00), 0.873 (95% CI: 0.711, 1.00), and 0.875 (95% CI: 0.688, 1.00), respectively. In the external validation set, the C-indexes of the three trained RSF models were higher than that of the Ki67 index (0.838, 0.754, and 0.866 vs. 0.601, respectively).
Random survival forest models developed using radiomics features derived from MRI images were proven helpful for accurate prediction of local recurrence of primary DFSP after surgical treatment and showed better predicting performance than the Ki67 index.
隆突性皮肤纤维肉瘤（DFSP）以局部侵袭性生长模式和高局部复发率为特征。准确识别局部复发风险高的患者可能有利于患者的随访，并对治疗决策具有潜在价值。本研究旨在探讨基于机器学习的放射组学模型是否能够准确预测原发性 DFSP 术后的局部复发。
本回顾性研究共纳入 2010 年至 2016 年期间来自两个不同机构的 146 例 DFSP 患者的 MRI 扫描结果：机构 1（n=104）用于训练集，机构 2（n=42）用于外部验证集。使用 MRI 图像开发了三个放射组学随机生存森林（RSF）模型。此外，还在外部验证集中比较了 Ki67 指数与三个 RSF 模型的性能。
训练集中，基于脂肪抑制 T2W（FS-T2W）图像、脂肪抑制 T1W 加钆对比（FS-T1W+C）图像和 FS-T2W 和 FS-T1W+C 联合图像的 10 折交叉验证中，RSF 模型的平均一致性指数（C-index）评分分别为 0.855（95%CI：0.629，1.00）、0.873（95%CI：0.711，1.00）和 0.875（95%CI：0.688，1.00）。在外部验证集中，三个训练的 RSF 模型的 C 指数均高于 Ki67 指数（0.838、0.754 和 0.866 与 0.601 相比）。
使用 MRI 图像提取的放射组学特征开发的随机生存森林模型有助于准确预测原发性 DFSP 术后的局部复发，其预测性能优于 Ki67 指数。
