Surgical videos of pituitary surgery can be uploaded in this webapp to get bounding box predictions of 15 different anatomical structures and the detection of surgical instruments. The detection model is the standard implementation of YOLOv7. The model was trained, validated, and tested on a dataset consisting of 166 pituitary surgery videos (transsphenoidal adenomectomy), where 20 of those were used for testing. The entire dataset contains approximately 19000 labeled frames. 

To enable further research on this topic we provide the weights of the model as well as predictions and confidences for the uploaded videos. Uploaded videos will be deleted 1 week after processing. 

In case you use this tool for your research, please consider citing our work:

https://doi.org/10.1007/978-3-031-34048-2_9
https://arxiv.org/abs/2405.09355

This work was funded by the SNSF.