Extended color code and Automated Quantification of Islet cells

Alpha (α) and Beta (β) Islet cells in pancreas of Langerhans can be utilized for analyzing primitive diabetes of both types 1 and 2. Manual counting and conventional investigation of such histological tissues is quite laborious, time taking, less reliable, and subject to error. In order to achieve more accurate results as quickly as possible, the presented work devices a novel color scheme code (I4) from existing schemes (I1,I2,I3). Moreover the work offers an Automated Quantification technique for enumeration and segmentation of pancreatic mass from microscopic images. The technique includes five modules; (1) Segmentation to magnify Islet cells from the rest of mass, (2) Equalization uses Contrast-Limited Adaptive Histogram equalization method in order to enhance contrasts between different intensity levels, (3) Color Space Formation uses existing color spaces with proposed smart calculations to construct novel color space code, (4) Image Conversion applies novel color space code on the image followed by certain post processing (binarization, hole-filling, erosion), and (5) Cell Analysis uses connected component labeling to finally identify and label the cells. The technique is evaluated against a number of diverse microscopic images obtaining satisfactory empirical results with recognition confidence interval in terms of true positive, false positive, and false negative are respectively 97.75 ± 2.64, 1.79 ± 1.04, and 1.37 ± 0.84. The work offers two contributions: it devices a novel color scheme, and offers an automated quantification methodology.