ABSTRACT
Human health is the real wealth for a society. Consequently prevention of health from complex diseases like
cancer needs the diagnosis of these entire viruses at an early stage. Colon cancer, the most common one, reached
the highest rate among all the other types recently. Colorectal cancer gets developed either in colon or in the
rectum inside the large intestine, due to the abnormal growth of the cells. Computer-aided decision support
system has become one of the major research topics in medical imaging field during the past two decades to detect
cancers. Detecting and screening of colorectal cancers are done by a Computed Tomography. The implemented
algorithm determines the locations and features of glands which are affected by cancer tissues and save this
information for the subsequent diagnosis. The proposed algorithm carries out the diagnosis with two modules:
One known as the gland detection and the other one referred as the nuclei detection. Gland detection is performed
in the proposed algorithm using color segmentation either through HSV or LAB transformation. Noise removal
and erosion of the input image is performed for enhancing the selection of the affected tissues. The boundary
detection and connection is established through Markov Chain model to identify the affected tissues with proper
threshold. The first module detects the glands where the possibly of miss detection is more. Hence to remove the
miss detected glands the algorithm proceed for the second module referred as nuclei detection. The most well
known region growing methodology is slightly modified to increase the speed and reduce the memory size To
provide the execution in low-end clients, the whole image is cracked into smaller tiles and after the processing of
each individual tiles , the results are to be merged to get back the original size. After nuclei detection if the
number of nucleus is more that glands are miss detected glands and they are removed
Keywords: - Gland detection, colon cancer, nuclei detection