ABSTRACT
Segmenting abdominal organs remains a critical component both in the provision of medical diagnosis and in planning for treatment. In this regard, the present work introduces articles “MedSolution”, which is a system composed of segmentation and classification components withsin a single system aimed at aiding in the determination of the appropriateness of surgery for the patients with abnormality of abdominal organs. MedSolution uses a U-Net-based model to segment the abdominal organs from the medical images. After segmenting the organs, it assesses the size of the organ and patient survival information by using a voting classifier that aggregates support vector machine, Random forests, and XGBoost, to estimate the probability of surgical intervention. Considering both organ size metrics and classification results, MedSolution gives a data-driven recommendation to clinicians on the need for surgical intervention. Such approach aims to improve the accuracy of diagnosis, the efficiency of clinical processes and health outcomes of patients and as such acts as a great benefit in the early recognition of diseases that may need surgical treatment and in aiding clinical decisions more efficiently.
