Human brain tumor diagnosis using the combination of the complexity measures and texture features through magnetic resonance image. (August 2020)

• Record Type: Journal Article
• Authors: Salem Ghahfarrokhi, Sepehr; Khodadadi, Hamed
• Is Part Of: Biomedical signal processing and control. Volume 61(2020)
• Publication Date: 2020-08
• Subjects: Brain tumors -- CAD -- DWT -- GLCM -- Complexity measures -- Pattern net -- MRI; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2020.102025
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: In this paper, several approaches for MRI classification are addressed. Complexity indices, GLCM based on DWT and their combination are the used techniques. Comparison with the other similar works showed the efficiency of the proposed system. Abstract: The brain tumor is known as the main reason for death. Hence, knowing the type of brain tumors plays an important role in diagnosis and treatment. Traditional invasive methods like a biopsy, lumbar puncture, and spinal tap have been employed for the detection and classification of these tumors. In this paper, a Computer-Aided Diagnosis (CAD) system is provided for the classification of these tumors in Magnetic Resonance Imaging (MRI). For this purpose, the chaos theory is utilized for estimating the complexity measures such as Lyapunov Exponent (LE), Approximate Entropy (ApEn), and Fractal Dimension (FD). Furthermore, by extraction of Gray-Level Co-occurrence Matrix (GLCM) and Discrete Wavelet Transform (DWT)-based features, the benign and malignant tumors could be distinguished. The calculated features are applied to three classifiers such as Support Vector Machine (SVM), K-Nearest Neighbors (KNN) algorithm, and pattern net. In the validation step, several experiments are carried out on the various combinations of features and classifiers. Accordingly, the best accuracy (98.9%) is attained by incorporating complexity measures with GLCM features and pattern net classifier. Also, the comparison between the results of this study and other similar works with the same dataset demonstrates the efficiency of the proposed method.