Abstract:
Apparent Diffusion Coefficient (ADC) is one of the most common magnetic
resonance imaging (MRI) techniques that are frequently used in the brain tumor
diagnosis process. This study is based on extracting statistical texture features
from MRI-ADC images of human brain tumors to observe correlations of feature
values between malignant and benign brain tumors. This study was carried out
using 980-malignant, 813-benign labeled MRI brain ADC image slices acquired
from 253 subjects presented to the National Hospital of Sri Lanka. The
pathological condition of each subject was identified by the radiological reports
and confirmed it using histopathological reports. Pixel values within the tumor
region of the selected ADC images were delineated by drawing region of interest
(ROI) surrounding the tumor area. The features; mean pixel value, higher-order
moments of ADC, Grey Level Co-occurrence Matrix (GLCM) texture features;
mean, variance, energy, entropy, contrast, homogeneity, correlation, prominence
and shade, and patients’ age were extracted from each ROI. The extracted features
were tested with a one-tailed P-value test with a 95% confidence level. The values
for kurtosis of ADC, mean pixel value of ADC, patient age, and the GLCM
texture features; mean1, mean2, variance1, variance2, energy, and contrast
showed significantly (P-value<0.05) higher feature values for benign tumors
while the entropy, homogeneity correlation, prominence, and shade showing
significantly high values for malignant tumors. The facts for the skewness of ADC
were not enough (P-value 0.05<0.0734) to reject the null hypothesis. The study
concludes the feasibility of utilizing the above features except for skewness, as
potential biomarkers to differentiate benign and malignant brain tumors.