AI-Assisted MRI: Navigating Deep Endometriosis

Artificial Intelligence, Genital / Reproductive system female, MR, Computer Applications-3D, Computer Applications-Detection, diagnosis, Computer Applications-General, Image verification Authors: A. Dafinescu, R. M. Manea, R. E. Birla Coroiu DOI: 10.26044/ecr2026/C-27710 Learning objectives Review the key MRI features of deep infiltrating endometriosis (DIE) across different pelvic compartments. Explain the basic principles of artificial intelligence (AI) relevant to pelvic MRI interpretation, including detection, segmentation and radiomics. Demonstrate how AI-assisted tools can complement conventional MRI interpretation in DIE. Discuss current limitations, pitfalls, sources of error and clinical considerations associated with AI use in this setting.

Deep infiltrating endometriosis (DIE) is the aggressive form of endometriosis characterized by fibromuscular infiltration extending more than 5 mm beneath the peritoneal surface. Commonly affects multiple pelvic compartments, including the uterosacral ligaments, posterior vaginal fornix, rectovaginal septum, bowel (rectosigmoid), bladder, parametria, pelvic sidewalls and pelvic nerves [1]. DIE is frequently multifocal and associated with dense adhesions that distort normal pelvic anatomy. Accurate preoperative assessment is essential for optimal surgical planning and reduction of postoperative complications [1,2].Magnetic resonance imaging (MRI) is the reference imaging modality for... Findings and procedure details MRI Protocol and Imaging FeaturesAccurate evaluation of DIE requires a dedicated pelvic MRI protocol. High-resolution multiplanar T2-weighted sequences are essential for pelvic anatomy and fibrotic lesions. Axial and sagittal T1-weighted images, with or without fat suppression, help identify hemorrhagic foci. Oblique planes improve visualization of specific structures, particularly the uterosacral ligaments. Diffusion-weighted imaging has an adjunctive role in selected cases, mainly for differential diagnosis and exclusion of malignancy. Contrast-enhanced sequences are not routinely required but may be useful in specific situations. Standardized acquisition, appropriate patient...

DIE is a complex, multifocal disease requiring detailed preoperative imaging assessment, thus MRI remains the cornerstone imaging modality for diagnosis and mapping. New developed AI-assisted tools can support detection, segmentation and compartment-based mapping of disease, but AI outputs must be critically evaluated and integrated with conventional MRI findings, so as a consequence the radiologist expertise and oversight remain essential for safe and effective clinical use of AI. Personal information and conflict of interest A. Dafinescu: Nothing to disclose R. M. Manea: Nothing to disclose R. E. Birla Coroiu: Nothing to disclose

1. Rousset P, Florin M, Bharwani N, Touboul C, Monroc M, Golfier F, et al. Deep pelvic infiltrating endometriosis: MRI consensus lexicon and compartment-based approach from the ENDOVALIRM group. Diagnostic and Interventional Imaging. 2022 Nov;104(3). https://doi.org/10.1016/j.diii.2022.09.0042. Lorusso F, Scioscia M, Rubini D, Stabile Ianora AA, Scardigno D, Leuci C, et al. Magnetic resonance imaging for deep infiltrating endometriosis: current concepts, imaging technique and key findings. Insights into Imaging. 2021 Jul 22;12(1). https://doi.org/10.1186/s13244-021-01054-x3. Bazot M, Daraï E. Diagnosis of deep endometriosis: clinical examination, ultrasonography, magnetic resonance...