Improvement in automated diagnosis of liposarcoma using machine learning

Computer engineering

Sarcomas in the form of soft tissue tumours (STT) can develop in the connective, encircling, and supporting tissues of the body. When seen by Magnetic Resonance Imaging, they appear to be heterogeneous due to their shallow frequency in the body and their great diversity. They are frequently confused with other illnesses such lymphadenopathy, struma nodosa, and fibro adenoma mammae, and these diagnostic mistakes have a significant negative impact on the medical treatment of patients. Numerous machine learning models have been put forth by researchers to categorise cancers, but none have sufficiently addressed the issue of incorrect diagnoses. Additionally, comparable studies that have suggested methods for evaluating these tumours typically do not take the heterogeneity and magnitude of the data into account. For this reason, we suggest a machine learning-based strategy that combines a novel method of preprocessing the data for feature transformation, resampling methods to remove bias and the deviation of instability, and performing classifier tests using the Support Vector Machine (SVM) and Logistic Regression algorithms (LR). Tests conducted on data gathered at Yogyakarta, Indonesia's Nur Hidayah Hospital, reveal a significant advancement over earlier research. These findings support the idea that machine learning techniques could offer practical and useful tools to support STT diagnostics' automatic decision-making procedures

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.