Brain Tumor Classification Using Gray Level Co-occurrence Matrix and Convolutional Neural Network

Abstract

Image are objects that have many information. Gray Level Co-occurrence Matrix is one of many ways to extract information from image objects. Wherein, the extracted informations can be processed again using different methods, Gray Level Co-occurrence Matrix is use for clarifying brain tumor using Convolutional Neural Network. The scope in this research is to process the extracted information from Gray Level Co-occurrence Matrix to Convolutional Neural Network where it will processed as Deep Learning to measure the accuracy using four data combination from TI1, in the form of brain tumor data Meningioma, Glioma and Pituitary Tumor. Based on the implementation of this research, the classification result of Convolutional Neural Network shows that the contrast feature from Gray Level Co-occurrence Matrix can increase the accuracy level up to twenty percent than the other features. This extraction feature is also accelerate the classification process using Convolutional Neural Network.

Brain tumor detection from MRI images with using proposed deep learning model: the partial correlation-based channel selection

“…While the proposed CNN classifier achieved an accuracy rate of 98.93% and a sensitivity rate of 98.18% for cropped lesions, for uncropped it showed a 99% accuracy and 98.52% sensitivity rate, for segmented lesions its accuracy rate was 97.62%. Widhiarso et al [24], in 2018; presented a brain tumor classification model using CNN and GLCM (Gray Level Co-Occurrence Matrix). The study extracted energy, correlation, contrast, and homogeneity features from four different angles for each image.…”

Klasifikasi Tumor Otak pada Citra Magnetic Resonance Image dengan Menggunakan Metode Support Vector Machine

“…GLCM merupakan suatu metode untuk melakukan ekstraksi ciri berbasis statistikal, perolehan ciri diperoleh dari nilai matriks yang mempunyai nilai tertentu dan membentuk sudut pola. Untuk sudut yang dibentuk dari nilai piksel citra menggunakan GLCM adalah 0º, 45º, 90º, 135º ( [1], [2]). Dari piksel-piksel tersebut terbentuk matrix co-occurance dengan pasangan pikselnya.…”

Brain Tumor Classification Using Deep Learning Technique - A Comparison between Cropped, Uncropped, and Segmented Lesion Images with Different Sizes

“…The results show that they achieve an accuracy rate of 96.97% and a sensitivity of 97.0 %. Widhiarso, Wijang, Yohannes Yohannes, and Cendy Prakarsah [10] presented a brain tumor classification system using a convolutional neural network (CNN) and Gray Level Co-occurrence Matrix (GLCM) based features. They extracted four features (Energy, Correlation, Contrast, and Homogeneity) from four angles (0°, 45°, 90°, and 135°) for each image and then these features are fed into CNN, they tested their methodology on four different datasets (Mg-Gl, Mg-Pt, Gl-Pt, and Mg-Gl-Pt) and the best accuracy achieved was82.27% for Gl-Pt dataset using two sets of features; contrast with homogeneity and contrast with correlation.…”

“…However, if the physician suspects a brain tumor and they require more information about its type, a surgical biopsy from the suspected tissue (tumor) is necessary for a detailed diagnosis by the specialist. These different technologies in brain tissue imaging have increased over recent years for image contrast and resolution enhancement which allows the radiologist to identify even small lesions and therefore achieving higher diagnosis accuracy [7,8,9,10]. Combining (fusing) the images acquired from different imaging modalities and benefiting the recent advances of engineering technologies that enhance the accuracy of brain tumor detection in the field of artificial intelligence (AI) for computer vision applications, where AI can be integrated with these imaging modalities to build a computer-aided diagnosis (CAD) systems.…”