BPML, MLops and 3D-UNet Network Integration in End-to-End Application Design Applied to the Segmentation of Human Brain Tumors in Clinic Cases

This work aims to facilitate the integration of the scientific findings obtained from the BraTS challenge with business processes and applicable in real clinical cases. We describe the 3D-UNet neural network originally proposed by Ronneberger et al. in 2015 for biomedical image segmentation. We evaluate its performance for segmenting human brain tumors from real MRI-scan images data BraTS 2023 challenge. We very briefly explain the details of how these MRI scans are interpreted and the neural network architecture of the model obtained. Finally, we present the entire end-to-end BPM business integration through MLops for a clinical case of segmentation of brain tumors. We conclude that the proposed methodology allows the creation of an end-to-end operational proof of concept app in a business logic agilely based on a segmentation model to value the research results.