Unsupervised Multi-Stage Segmentation in Image Identification of Gallstone Disease

Detection of gallstone disease on medical images plays a very active role in early screening. However, medical imaging still cannot provide optimal diagnostic results. This is because medical images have deficiencies in the identification process. With these deficiencies, it is necessary to process image processing to overcome existing problems. This study aims to identify gallstone disease using an image segmentation process. The segmentation process is carried out by adopting the performance of the Multistage Segmentation Algorithm (MSA) based on an unsupervised approach and image morphology techniques. The performance of this MSA is expected to obtain the results of identifying images of objects containing gallstones. The data set for this study consisted of 1678 images derived from the examination results of 17 gallstone patients based on medical imagery resonance cholangiopancreatography (MRCP). Based on the experiments conducted, MSA can present precise and accurate identification results. These results were obtained based on the performance of the Cluster-Based Segmentation (CBS) process and morphological image operations can to detected gallstone objects. The overall results of the segmentation process provide fairly good test results with an accuracy value of 90% With these results, this study contributes to optimizing the process of diagnosing gallstone disease patients.