Xuanxuan Li1, Yiping Lu2, Jianxun Qu3, Bo Yin2, and Daoying Geng2
1Radiology, Huashan Hospital Affiliated to Fudan University, Shanghai, China, 2Huashan Hospital Affiliated to Fudan University, Shanghai, China, 3Department of MR Research, GE Healthcare, Shanghai, China
Synopsis
We attempted 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). Our sample population consisted of 23 malignant HPCs and 43 AM. We compared the diagnostic ability of three classifiers based on texture features extracted from each modality (T2FLAIR, T1-CE, and DWI) to the classifier based on clinical features from three neuro-radiologists. The T1W-CE classifier performed the best.
Purpose
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).
Materials and Methods
Sixty-six pathologically confirmed cases, including 23 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 neuro-radiologists 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 neuro-radiologists were evaluated by receiver operating characteristic (ROC) analysis.Results
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 neuro-radiologists (AUC = 0.69, 0.70 and 0.73). Conclusion
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.Acknowledgements
No acknowledgement found.References
[1] Louis, D.N., et al., The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary. Acta Neuropathol, 2016. 131(6): p. 803-20.
[2] Liu, L., et al., Comparison of ADC values of intracranial hemangiopericytomas and angiomatous and anaplastic meningiomas. J Neuroradiol, 2014. 41(3): p. 188-94.
[3] Fatima, K., A. Arooj, and H. Majeed, A new texture and shape based technique for improving meningioma classification. Microsc Res Tech, 2014. 77(11): p. 862-73.
[4] Kassner, A. and R.E. Thornhill, Texture analysis: a review of neurologic MR imaging applications. AJNR Am J Neuroradiol, 2010. 31(5): p. 809-16.
[5] Szczypinski, P.M., et al., MaZda--a software package for image texture analysis. Comput Methods Programs Biomed, 2009. 94(1): p. 66-76.