Global Journal of Science Frontier Research, A: Physics and Space Science, Volume 22 Issue 1

III. R esults and D iscussions The proposed methodology has been evaluated on images collected from the database that belongs to Mammocare Ghana, emphasizing the importance of K- Means clustering algorithms in cancerous mammography segmentation. In the proposed methodology, specifically, the effectiveness of the segmentation methods was evaluated on the RGB image, based on the intensity levels of the segmented output. The result of clustering intensities of the colour bands into different groups using the K-means clustering algorithm were a set of three distinct RGB level regions. These regions, referring to the colors existing in the original image are presented in figure 4, samples 1 to 8. Each region is relatively homogeneous in terms of pixel intensity. These regions were breast intensities (cluster 1), background intensities (cluster 2) and tumor intensity (cluster3) referring to the colors existing in the original image[18]. Therefore, it was assumed that there were three classes of objects to be separated with the K-means clustering algorithm. The figures show an original image from the image database and results for clusters using the K- means clustering method with only 4 clusters, and varying the values of classes are shown. Four clusters were used because using three clusters was not sufficient in that case due to the natural variability of sharpness in the input mammogram image. It can be seen that mass and lesion elements in the breast image became clearer by increasing the number of classes keeping the constant value of bins, visual appearance and classification of microcalcification get improved[6]. The following figures show the original images and their processed images using K-means clustering in L*a*b* colour spaces. (A) (B) (C) (D) Figure 4: Sample 1 shows, (A) the original image, (B) cluster 1, (C) cluster 2 and (D) cluster 3 after separating objects by colour using K-means clustering technique in L*a*b* colour space with the tumor region circled in yellow (A) (B) (C) (D) Figure 5: Sample 2 shows, (A) the original image, (B) cluster 1, (C) cluster 2 and (D) cluster 3 after separating objects by colour using K-means clustering technique in L*a*b* colour space with the tumor region circled in yellow Segmentation of Cancerous Mammography using MATLAB 1 Year 2022 47 © 2022 Global Journals Global Journal of Science Frontier Research Volume XXII Issue ersion I VI ( A )