Department of Radiology, The Affiliated Hospital of Qingdao University Qingdao, 16 Jiangsu Road, Qingdao, Shandong, China.
Department of Pathology, The Affiliated Hospital of Qingdao University, Qingdao, Shandong, China.
Cancer Imaging. 2021 Feb 6;21(1):20. doi: 10.1186/s40644-021-00387-6.
We sought to evaluate the performance of a computed tomography (CT)-based radiomics nomogram we devised in distinguishing benign from malignant bone tumours.
Two hundred and six patients with bone tumours were spilt into two groups: a training set (n = 155) and a validation set (n = 51). A feature extraction process based on 3D Slicer software was used to extract the radiomics features from unenhanced CT images, and least absolute shrinkage and selection operator logistic regression was used to calculate the radiomic score to generate a radiomics signature. A clinical model comprised demographics and CT features. A radiomics nomogram combined with the clinical model and the radiomics signature was constructed. The performance of the three models was comprehensively evaluated from three aspects: identification ability, accuracy, and clinical value, allowing for generation of an optimal prediction model.
The radiomics nomogram comprised clinical and radiomics signature features. The nomogram model displayed good performance in training and validation sets with areas under the curve of 0.917 and 0.823, respectively. The areas under the curve, decision curve analysis, and net reclassification improvement showed that the radiomics nomogram model could obtain better diagnostic performance than the clinical model and achieve greater clinical net benefits than the clinical and radiomics signature models alone.
We constructed a combined nomogram comprising a clinical model and radiomics signature as a noninvasive preoperative prediction method to distinguish between benign and malignant bone tumours and assist treatment planning.
我们旨在评估我们设计的基于计算机断层扫描（CT）的放射组学列线图在区分良性和恶性骨肿瘤方面的性能。
将 206 名骨肿瘤患者分为两组：训练集（n=155）和验证集（n=51）。使用基于 3D Slicer 软件的特征提取过程从增强 CT 图像中提取放射组学特征，并用最小绝对值收缩和选择算子逻辑回归计算放射组学评分以生成放射组学特征。临床模型包括人口统计学和 CT 特征。构建了一个包含临床模型和放射组学特征的放射组学列线图。从识别能力、准确性和临床价值三个方面综合评估了三个模型的性能，以生成最佳预测模型。
放射组学列线图包含临床和放射组学特征。该列线图模型在训练集和验证集的表现良好，曲线下面积分别为 0.917 和 0.823。曲线下面积、决策曲线分析和净重新分类改善表明，放射组学列线图模型可以获得比临床模型更好的诊断性能，并比临床和放射组学特征模型单独使用获得更大的临床净效益。
我们构建了一个包含临床模型和放射组学特征的联合列线图，作为一种非侵入性的术前预测方法，用于区分良性和恶性骨肿瘤并辅助治疗计划。
