To develop a computer vision tool that can automatically confirm the diagnosis and assessment of the different endometriosis lesions is the aim for WP6.
Here’s a selected few key FEMaLe outputs and highlights in WP6 from July 2022 to December 2023:
– We have demonstrated that applying deep learning techniques in medical image analysis is possible even with very diverse data, such as lesion images from laparoscopic video. One of the main aspects is addressing the challenges of cross-border image sharing, especially considering stringent regulations like GDPR.
Results from WP6 were presented during the World Congress on Endometriosis in Edinburgh in May, 2023:
* Applying Cross-Border Machine Learning Approach for GDPR-protected Medical Data (Poster)
– A significant accomplishment is the development of efficient strategies for data preparation and dataset splitting. We have effectively mitigated the issue of data leakage between training and testing phases, ensuring the robustness of our models while preserving stratification – classes balance between train, test and validation sets.
– The successful integration of the algorithm into the Augmented Reality (AR) software marks a significant milestone for FEMaLe and for SurgAR company. Our algorithm, previously verified for its commendable frame rate of 40 frames per second during inference, has now been seamlessly incorporated into the AR software. This integration is the result of dedicated efforts by our skilled development team.
