Artificial inTelligence as Tool for Early Diagnosis and Precision Surgery in eNdometriosis-related ovArian Cancer

Endometriosis (EMS) is a chronic, invaliding, inflammatory gynaecological condition affecting 10-15% of women in reproductive age. EMS is characterized by lesions of endometrial-like tissue outside the uterus involving pelvic peritoneum and ovaries. In addition, distant foci are sometimes observed. Unfortunately, the aetiology of the EMS is little known. Although non-malignant, EMS shares similar features with cancer, such as development of local and distant foci, resistance to apoptosis and invasion of other tissues with subsequent damage to the target organs. Moreover, patients with EMS (particularly ovarian EMS) showed high risk (about 3 to 10 times) of developing epithelial ovarian cancer (EOC). Epidemiologic, morphological and molecular studies reported endometrioma as the precursor of EOC, including clear cell (CCC) endometrioid carcinoma which are both called "EMS-related ovarian carcinoma (EROC)". To date, it remains unclear why benign EMS causes malignant transformation. This multi-step process, unlike high-grade serous carcinomas, offers the possibility to identify the carcinoma precursors enabling an early diagnosis and in the early stages of the disease.

EOC is the most lethal female gynecological cancer with 25% 5-year overall survival (OS), due to the lack of effective screening tools, and rapidly spreads over the entire peritoneal surface (carcinosis) thus involving all abdominal organs. Diagnosis and clinical staging of EOC is currently performed by qualitative image evaluation although the sensitivity/specificity is suboptimal. To date, diagnostic, staging, and prognostic factors are strongly correlated with subjective assessment training and clinician experience.

Genomic analysis based on Next Generation Sequencing (NGS) has revealed the presence of cancer-associated gene mutations in EMS. Moreover, the chronic inflammatory process of EMS involves many factors, such as hormones, cytokines, glycoproteins, and angiogenic factors, which are expected to become early EMS biomarkers.

A promising new branch of cancer research is the use of artificial intelligence (AI) to recognize new image patterns and texture and/or detecting novel biomarkers to improve the early identification of EROC patients. AI has never been used for EROC and we want to investigate whether these methods/techniques can support and even improve current diagnostics and risk assessment. AI will be used to construct a new 3D risk assessment model based on images and volume of interest