MORPHOLOGICAL SEGMENTATION OF THE KIDNEYS FROM ABDOMINAL CT IMAGES

AICHA BELGHERBI, ISMAHEN HADJIDJ, ABDELHAFID BESSAID
2014 Journal of Mechanics in Medicine and Biology  
The liver is a common site for the tumors occurrence. Automatic hepatic lesion segmentation is a crucial step for diagnosis and surgery procedure. When computed tomography (CT) scans is used, liver lesions segmentation is a challenging task. In CT images, hepatic lesions located in a liver are generally identified by intensity difference between lesions and liver. In fact, the intensity of the lesions can be lower or higher than that of the liver. In this work, the segmentation is performed in
more » ... wo stages. In the primary stage, the liver structure will be at first extracted from the image using the morphological reconstruction. The second stage is devoted to detect the hepatic lesions by the watershed transform. However, the segmentation procedure; as it is mentioned previously; is a difficult task. In fact, the main problem of liver lesions detection from CT images is related to low contrast between lesions and liver intensities. To solve this problem, a new method developed for the semi-automatic segmentation of hepatic lesions is developed. It is based on the anatomical information and mathematical morphology tools that are used in the image processing field. The proposed segmentation algorithm is evaluated by comparing our results with the manual segmentation performed by an expert. post-processing operation to achieve an automatic segmentation of liver. Then, they used an alternative Fuzzy C-Means clustering in order to get the lesion segmentation. JOURNAL OF BIOMEDICAL SCIENCES Conclusion The presented system for detection and segmentation of focal liver lesions is able to reliably segment the lesions in the used patient database. This method can be helpful for preliminary testing of CT scan images. Our proposed approach has been tested for many other images and it gives satisfactory results. The results of segmentation were 92% of detection rate and 99 % of specificity. In the future the segmentation may be improved by treating differently and multiple lesion also the boundaries of the liver.
doi:10.1142/s0219519414500730 fatcat:uvawsx7h65cgvd2wk5337527ia