Detection of Ovarian Tumors Using TVUS Imaging and Segmentation Techniques

Abstract

Listen

Translate article

Ovarian most cancers remains a giant health subject due to its regularly asymptomatic development until superior stages, where treatment options are confined. Early detection is critical for enhancing affected person outcomes and survival costs. This observe proposes a comprehensive technique leveraging superior photograph processing techniques for the early detection and category of ovarian tumors the usage of trans vaginal ultrasound (TVUS) snap shots. The method starts with preprocessing steps which include median filtering to do away with noise artifacts inclusive of salt and pepper noise from ultrasound photos. Ultimately, picture enhancement techniques are applied to enhance contrast and intensity, observed through binary sample evaluation to signify texture features vital for tumor identification. Photo segmentation is performed the use of okay- manner clustering and gray degree Co-prevalence Matrix (GLCM) analysis to isolate and analyze areas of interest within the ovarian photos. A Convolutional Neural community (CNN) is then hired for correct classification of ovarian tumors as regular, benign, or malignant primarily based on the extracted functions. This approach not handiest enhances the diagnostic accuracy but also minimizes affected person publicity to ionizing radiation, making it appropriate for ordinary medical applications. The proposed methodology represents a considerable advancement within the discipline of ovarian cancer detection, presenting a non-invasive and green way to evaluate ovarian health and facilitate timely intervention. Destiny research directions include refining the set of rules’s performance with large datasets and exploring additional imaging modalities for complete ovarian fitness evaluation.