Presurgical differentiation between malignant haemangiopericytoma and angiomatous meningioma by a radiomics approach based on texture analysis

• Published in: Journal of neuroradiology Vol. 46; no. 5; pp. 281 - 287
• Main Authors: Li, Xuanxuan, Lu, Yiping, Xiong, Ji, Wang, Dongdong, She, Dejun, Kuai, Xinping, Geng, Daoying, Yin, Bo
• Format: Journal Article
• Language: English
• Published: France Elsevier Masson SAS 01.09.2019
• Subjects: Adult; Diagnosis, Computer-Assisted - methods; Female; Haemangiopericytoma; Hemangiopericytoma - diagnostic imaging; Hemangiopericytoma - pathology; Humans; Machine Learning; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; Male; Meningeal Neoplasms - diagnostic imaging; Meningeal Neoplasms - pathology; Meningioma; Meningioma - diagnostic imaging; Meningioma - pathology; Middle Aged; Neuroradiography - methods; Preoperative Care; Retrospective Studies; Sensitivity and Specificity; Support vector machine; Support vector machine; Magnetic resonance imaging; Haemangiopericytoma; Machine-learning; Meningioma
• ISSN: 0150-9861
• DOI: 10.1016/j.neurad.2019.05.013

[Display omitted] •A precise preoperative differentiation between HPC and AM is important for further treatment plans.•HPCs were evaluated to have larger sizes, more serpentine-like intratumoural vessels, and slighter degrees of peritumoural oedema compared with AMs.•Machine-learning models based on clinical features alone did not outperform the experienced neuroradiologists.•SVM classifiers based on texture features extracted from conventional MRI sequences can achieve a better diagnostic performance than classifiers built on clinical features, and enhanced T1WI is the best sequence for TA. To assess whether a machine-learning model based on texture analysis (TA) could yield a more accurate diagnosis in differentiating malignant haemangiopericytoma (HPC) from angiomatous meningioma (AM). Sixty-seven pathologically confirmed cases, including 24 malignant HPCs and 43 AMs between May 2013 and September 2017 were retrospectively reviewed. In each case, 498 radiomic features, including 12 clinical features and 486 texture features from MRI sequences (T2-FLAIR, DWI and enhanced T1WI), were extracted. Three neuroradiologists independently made diagnoses by vision. Four Support Vector Machine (SVM) classifiers were built, one based on clinical features and three based on texture features from three MRI sequences after feature selection. The diagnostic abilities of these classifiers and three neuroradiologists were evaluated by receiver operating characteristic (ROC) analysis. Malignant HPCs were found to have larger sizes, slighter degrees of peritumoural oedema compared with AMs (P<0.05), and more serpentine-like vessels. The AUC of the enhanced T1WI-based classifier was 0.90, significantly higher than that of T2-FLAIR-based or DWI-based classifiers (0.77 and 0.73). The AUC of the SVM classifier based on clinical features was 0.66, slightly but not significantly lower than the performances of 3 neuroradiologists (AUC=0.69, 0.70 and 0.73). Machine-learning models based on clinical features alone could not provide a better diagnostic performance than that of radiologists. The SVM classifier built by texture features extracted from enhanced T1WI is a promising tool to differentiate malignant HPC from AM before surgery.