Cooccurrence Histograms for characterizing brain magnetic resonance images

Cooccurrence Histogram (COH) is a graphical representation derived from cooccurrence matrix as a form of plot between the joint probability entries of the direction independent matrix, and their percentage frequencies. The advantage of COH representation is that similarities and dissimilarities of different textures can be perceived as all matrix information is visually displayed. Applying COH on brain tumors and surrounding white matter showed different histogram peaks for tumoral and nontumoral regions. COH for tumoral regions seemed to be more flattened with numerous peaks. Normal white matter showed three narrow peaks at distinct locations. Skewness and Kurtosis of COH were calculated and compared. Different peaks of COH might indicate the presence of different texture components within the region, and can be used to reveal subtle modifications to the tissue if their overall shape or number changes.