Introduction

The differential diagnosis model of uterine sarcomas is based on a machine learning classifier, trained and tested for binary classification task: malignant vs benign, based on histopathology results collected on a dataset of patients.

Technical features

The current major issue is the difficulty in making an accurate and fast diagnosis for patients with SU and choosing the right surgical approach not only for the patient’s prognosis but also for uterine preservation. Currently, the diagnosis of SU is almost always defined in a post-operative context. The preoperative histological sampling is challenging because the malignant lesion usually presents as heterogeneous with both benign and malignant areas. Furthermore, the persistent problem of neoplastic dissemination during biopsy remains.

The solution we are presenting is called SARA, which is a software with an intuitive graphical interface for the differential diagnosis of benign (MU) and malignant (SU) lesions through radiomics. This will enable the radiologist to make an accurate and rapid diagnosis, regardless of their experience, without the need for extensive training and without the requirement for bioinformatics knowledge. The only prerequisite is identifying volumes of interest in contrast-enhanced preoperative CT images.

Possible Applications

• Biotechnology sector;
• Biomedical sector;
• Pharmaceutical sector;
• Therapeutics;
• Diagnostic field.

Advantages

• Differential diagnosis between US and UM;
• High diagnostic accuracy;
• Standardization of diagnosis;
• Personalization of therapeutic approach.