Our project aims to revolutionize the preparation process for brain surgery by leveraging advanced technologies. Currently, brain segmentation annotators undergo extensive training to manually label brain parts on a 3D model generated from brain MRI images. However, this training process is time-consuming, typically lasting 1-2 months, and still requires additional corrections. To address these challenges, the project proposes an innovative end-to-end solution that utilizes deep learning models to automate the segmentation of brain parts from MRI images and generate a comprehensive 3D model. The key differentiator is the compatibility of the final model with Virtual Reality (VR) tools, which empowers surgeons to immerse themselves in a virtual environment and conduct detailed investigations of the patient's brain.
We employ a well-defined methodology that encompasses data preparation, model training, and brain segmentation/3D model generation. By leveraging deep learning models and integrating them with VR tools, the project aims to automate the preparation process for brain surgery, enhance segmentation accuracy, and provide surgeons with advanced visualization capabilities. The ultimate goal is to improve the overall efficiency and success rates of brain surgeries by facilitating more accurate surgical planning and decision-making.
Source codes can be found in src directory.
Documents including resources and investigation documents can be found in documents directory.