| dc.description.abstract | Prostate cancer is one of the most common cancers in males and one of the
significant causes of cancer mortality. Most prostate malignancies are presently
diagnosed based on an increased PSA level, despite this biomarker having only
limited accuracy. Prostate cancer differs from most other cancers because it is
frequently multifocal and does not appear as a single spherical mass. The illness
progresses at different rates, and it is frequently asymptomatic until it has gone to
late stages. Multi-parametric MRI (mpMRI) has advanced dramatically in the last 20
years, as has the treatment of localised prostate cancer. As a result, this research
aims to develop an algorithm to identify features based on the Local Binary Pattern
(LBP) based histogram and Grey Level Run Length Matrix (GLRLM) characteristics
of mpMRI images, to improve detection rate and accuracy of prostate cancer
diagnosis. Local binary patterns are texture descriptors that have been effectively
employed as image descriptors in various applications. Images were gathered from
a public image database to complete this work. The operator is applied to the
selected region of interest (ROI) to generate the LBP image. Texture pattern
probability was summarised into a histogram, and second-order statistics were
obtained using the GLRLM operator. The statistical significance of the eleven
characteristics was determined using an independent two-sample t-test using four
features from the histogram and seven features from the GLRLM operator. The
suggested approach yielded three favourable outcomes in the research, which can
be utilised to identify malignant tumours from benign tumours. The positive results
include the first-order statistics standard deviation and kurtosis and the secondorder
statistic Run Length Non-uniformity (RLN). | en_US |
