Machine Learning Technique for Improvised and Automated Diagnosis of Soft Tissue Tumor

Soft Tissue Tumor (STT) are malignancies that are formed in the connective, supportive and surrounding tissues of the body. Because of their modest frequency throughout the person's body and enormous variation, they look diverse when seen via MRI Scans. These tumors are frequently misdiagnosed as Struma nodose, lymphadenopathy and fibrous adenoma mammie. The diagnostics error has substantial unfavorable influence on patient's healthcare. Several Machine Learning (ML) algorithms have been developed to diagnose cancers, but none have satisfactorily addressed the problem of misdiagnosis. Moreover, similar research that have presented models for evaluating such cancers typically ignore heterogeneity and data size. In response, this research study proposes a novel system of training technique that combines methods for data pre-processing records and for function transition. The process has different strategies to minimize bias and destabilizing variation based on classification tests. The technique is based on the datasets collected at Nur Hidayah Hospital in Yogyakarta, Indonesia, and shows a significant advancement over earlier studies. These findings show that ML approaches shall be used to improve the efficiency and efficacy of STT diagnostics self-driving decision-making procedures. The proposed Decision Tree method performs the classification with maximum accuracy of 98%, Area Under Curve of 94.4% and F1 score of 98.3%.