Efficient brain lesion segmentation using multi-modality tissue-based feature selection and support vector machines

• Published in: International journal for numerical methods in biomedical engineering 2013-09, Vol.29 (9), p.905-915
• Main Authors: Fiot, Jean-Baptiste, Cohen, Laurent D., Raniga, Parnesh, Fripp, Jurgen
• Format: Article
• Language: English
• Subjects: Brain; Brain - anatomy & histology; Brain - pathology; brain lesion; Brain Neoplasms - pathology; Classification; Computer Science; Feature extraction; Humans; Image Processing; Image Processing, Computer-Assisted - methods; Lesions; Machine Learning; Magnetic Resonance Imaging; Medical Imaging; Neuroimaging - methods; Patients; Post-processing; Reproducibility of Results; Segmentation; Support Vector Machine; Support vector machines
• ISSN: 2040-7939; 2040-7947
• DOI: 10.1002/cnm.2537

SUMMARYSupport vector machines (SVM) are machine learning techniques that have been used for segmentation and classification of medical images, including segmentation of white matter hyper‐intensities (WMH). Current approaches using SVM for WMH segmentation extract features from the brain and classify these followed by complex post‐processing steps to remove false positives. The method presented in this paper combines advanced pre‐processing, tissue‐based feature selection and SVM classification to obtain efficient and accurate WMH segmentation. Features from 125 patients, generated from up to four MR modalities [T1‐w, T2‐w, proton‐density and fluid attenuated inversion recovery(FLAIR)], differing neighbourhood sizes and the use of multi‐scale features were compared. We found that although using all four modalities gave the best overall classification (average Dice scores of 0.54  ±  0.12, 0.72  ±  0.06 and 0.82  ±  0.06 respectively for small, moderate and severe lesion loads); this was not significantly different (p = 0.50) from using just T1‐w and FLAIR sequences (Dice scores of 0.52  ±  0.13, 0.71  ±  0.08 and 0.81  ±  0.07). Furthermore, there was a negligible difference between using 5 × 5 × 5 and 3 × 3 × 3 features (p = 0.93). Finally, we show that careful consideration of features and pre‐processing techniques not only saves storage space and computation time but also leads to more efficient classification, which outperforms the one based on all features with post‐processing. Copyright © 2013 John Wiley & Sons, Ltd. This paper presents a white matter hyper‐intensity segmentation method that combines advanced pre‐processing, tissue‐based feature selection and supports vector machines classification. The proposed pipeline has been validated on a database of 125 patients with four magnetic resonance image modalities. The classification performance has been evaluated with regard to the relative input of each modality, the feature type, the impact of feature selection, and compared with other supervised algorithms.