Department of Radiology, Boston Medical Center, Boston University School of Medicine, USA; Research Center for Cooperative Projects, Hokkaido University Graduate School of Medicine, Japan.
Department of Radiology, Boston Medical Center, Boston University School of Medicine, USA.
Clin Radiol. 2021 Sep;76(9):711.e1-711.e7. doi: 10.1016/j.crad.2021.03.017. Epub 2021 Apr 30.
To investigate the value of machine learning-based multiparametric analysis using 2-[F]-fluoro-2-deoxy-d-glucose positron-emission tomography (FDG-PET) images to predict treatment outcome in patients with oral cavity squamous cell carcinoma (OCSCC).
Ninety-nine patients with OCSCC who received pretreatment integrated FDG-PET/computed tomography (CT) were included. They were divided into the training (66 patients) and validation (33 patients) cohorts. The diagnosis of local control or local failure was obtained from patient's medical records. Conventional FDG-PET parameters, including the maximum and mean standardised uptake values (SUVmax and SUVmean), metabolic tumour volume (MTV), and total lesion glycolysis (TLG), quantitative tumour morphological parameters, intratumoural histogram, and texture parameters, as well as T-stage and clinical stage, were evaluated by a machine learning analysis. The diagnostic ability of T-stage, clinical stage, and conventional FDG-PET parameters (SUVmax, SUVmean, MTV, and TLG) was also assessed separately.
In support-vector machine analysis of the training dataset, the final selected parameters were T-stage, SUVmax, TLG, morphological irregularity, entropy, and run-length non-uniformity. In the validation dataset, the diagnostic performance of the created algorithm was as follows: sensitivity 0.82, specificity 0.7, positive predictive value 0.86, negative predictive value 0.64, and accuracy 0.79. In a univariate analysis using conventional FDG-PET parameters, T-stage and clinical stage, diagnostic accuracy of each variable was revealed as follows: 0.61 in T-stage, 0.61 in clinical stage, 0.64 in SUVmax, 0.61 in SUVmean, 0.64 in MTV, and 0.7 in TLG.
A machine-learning-based approach to analysing FDG-PET images by multiparametric analysis might help predict local control or failure in patients with OCSCC.
研究基于机器学习的多参数分析使用 2-[F]-氟-2-脱氧-d-葡萄糖正电子发射断层扫描（FDG-PET）图像预测口腔鳞状细胞癌（OCSCC）患者治疗结果的价值。
纳入 99 例接受预处理整合 FDG-PET/计算机断层扫描（CT）的 OCSCC 患者。他们被分为训练（66 例）和验证（33 例）队列。局部控制或局部失败的诊断是从患者的病历中获得的。通过机器学习分析评估常规 FDG-PET 参数，包括最大和平均标准化摄取值（SUVmax 和 SUVmean）、代谢肿瘤体积（MTV）和总肿瘤糖酵解（TLG）、定量肿瘤形态参数、肿瘤内直方图和纹理参数，以及 T 期和临床分期。还分别评估了 T 期、临床分期和常规 FDG-PET 参数（SUVmax、SUVmean、MTV 和 TLG）的诊断能力。
在训练数据集的支持向量机分析中，最终选择的参数是 T 期、SUVmax、TLG、形态不规则、熵和游程非均匀性。在验证数据集，所创建算法的诊断性能如下：敏感性 0.82、特异性 0.7、阳性预测值 0.86、阴性预测值 0.64 和准确性 0.79。在使用常规 FDG-PET 参数的单变量分析中，T 期和临床分期的每个变量的诊断准确性如下：T 期为 0.61、临床分期为 0.61、SUVmax 为 0.64、SUVmean 为 0.61、MTV 为 0.64 和 TLG 为 0.7。
通过多参数分析对 FDG-PET 图像进行基于机器学习的分析方法可能有助于预测 OCSCC 患者的局部控制或失败。
