Descripción
|
The PituPhase65 dataset includes 65 endoscopic pituitary surgery videos, with a mean duration of 90 minutes, performed at the Hospital Universitario 12 de Octubre (Madrid, Spain). The surgeries have been labelled in eight different surgical phases, including a Phase 0, called "Outside the body", to account for moments in video when the endoscope was outside the patient. Informed consent was obtained from all patients for the publication of the videos.
SurgeryAI is a research project focused on the integration of artificial intelligence and digital tools in the care for neurosurgery patients. This dataset was built for the development of surgical phase recognition models that to support better surgical planning, real-time guidance, and workflow automation during surgery. Grant SurgeryAI funded by MICIU/AEI/10.13039/501100011033.
|
Publicación relacionada
| IsSupplementTo: González-Cebrián, Á., Bordonaba, S., Pascau, J., Paredes, I., Lagares, A., & De Toledo, P. (2025). Attention in surgical phase recognition for endoscopic pituitary surgery: Insights from real-world data. Computers In Biology And Medicine, 191, 110222.doi: https://doi.org/10.1016/j.compbiomed.2025.110222 |
Notas
| This dataset constitutes an improvement over a previous dataset published by the same research group, called "PituPhase-SurgeryAI", which contained 44 endoscopic pituitary surgery videos. Now, the PituPhase65 dataset includes the original 44 surgeries, along with 21 new surgeries, as well as revised annotation files. More specifically, this dataset has the following improvements: - The addition of a new surgical phase (P1, Middle Turbinate Resection) to the annotation files, which correspond to the opening step performed by the otorhinolaryngologist. - The inter-annotator reliability test has been updated, revising the annotation in those surgeries in which the level of agreement between both annotators was lower than 80%. |