Assessment of surgical tool inpainting techniques in epilepsy neurosurgery

Inpainting refers to the process of removing a foreground object from an image and seamlessly replacing it with consistent and contextually relevant background pixels. In the context of epilepsy neurosurgery, training videos are recorded using a microscope-mounted camera to document surgical procedures. However, surgical instruments frequently obstruct the camera’s line of sight, partially occluding brain tissue that is critical for assessing surgical techniques. This project aims to develop advanced inpainting techniques to remove surgical instruments from video recordings while preserving the anatomical integrity of the underlying brain structures. We will investigate state-of-the-art methods to ensure that the reconstructed regions maintain high visual fidelity without introducing distortions or artifacts. By integrating these techniques into our neurosurgery simulation platform, we aim to enhance the training and evaluation of surgical procedures. Additionally, this study will provide valuable insights into the feasibility of inpainting as a viable alternative within our current workflow, potentially improving the clarity and usability of surgical training videos.

Faculty Supervisor:

Houssem Gueziri

Student:

Partner:

Université de Carthage

Discipline:

Computer science

Sector:

Education

University:

Université TÉLUQ

Program: