Automated screening of MRI brain scanning using grey level statistics

Abstract

This paper describes the development of an algorithm for detecting and classifying MRI brain slices into normal and abnormal by relying on prior-knowledge, that the two hemispheres of a healthy brain have approximately a bilateral symmetry. We use the modified grey level co-occurrence matrix method to analyze and measure asymmetry between the two brain hemispheres. 21 co-occurrence statistics are used to discriminate the images. The experimental results demonstrate the efficacy of our proposed algorithm in detecting brain abnormality with high accuracy and low computational time. The dataset used in the experiment comprises 165 patients with 88 patients having different brain abnormalities whilst the remainder do not exhibit any detectable pathology. The algorithm was tested using a tenfold cross-validation technique with 100 repetitions to avoid the result depending on the sample order. The maximum accuracy achieved for the brain tumors detection was 97.8% using a Multi-Layer Perceptron Neural Network.

MRI brain scan classification using novel 3-D statistical features

“…We address the abovementioned shortcoming by proposing a new algorithm that is used to extract texture features from MRI brain scans in a threedimensional (3D). In addition, we compare the results of this study with the 2D approach developed in [6].…”

“…The overall flow chart of the proposed algorithm was explained in details in [6]. It starts with the data collection phase from the Iraqi hospital.…”

“…The same data set of MRI brain scans that was used in [6], is used in this study. The dataset was collected from one of Iraqi hospital, and includes 165 patients, they were collected from two different scanners.…”

“…Typically, it includes image resizing, image enhancement, standardizing intensity range, MSP detection and correction, and aligning the MRI slices into the same spatial location. These algorithms were developed and explained in details in [6,19].…”

“…The main objective of this study is to compare and evaluate the results reported in an early study by the authors [6] with the current results using three-dimensional feature extraction method in terms of accuracy and computational complexity. The proposed method is used to extract the texture features that are used to discriminate normal and pathological patients without any clinician's interception; hence reducing clinicians' examination and interpretation time.…”