Related Experiment Video
Updated: May 8, 2025
07:43
Endoscopic Endonasal Trans-sphenoidal Approach: Minimally Invasive Surgery for Pituitary Adenomas
Published on: January 17, 2018
18.8K
PitRSDNet: Predicting intra-operative remaining surgery duration in endoscopic pituitary surgery
Anjana Wijekoon1,2, Adrito Das1, Roxana R Herrera1
1UCL Hawkes Institute University College London London UK.
Healthcare Technology Letters
|December 25, 2024
Summary
Accurate prediction of remaining surgery duration (RSD) in pituitary surgery is crucial for patient care and cost efficiency. A new spatio-temporal neural network, PitRSDNet, effectively predicts RSD by learning from workflow sequences.
Area of Science:
- Medical Imaging and Artificial Intelligence
- Surgical Workflow Analysis
- Predictive Modeling in Healthcare
Background:
- Accurate intra-operative Remaining Surgery Duration (RSD) predictions are vital for anesthetic management and efficient operating room scheduling.
- Endoscopic pituitary surgery presents unique challenges for RSD prediction due to variable workflow sequences and optional surgical steps.
- Existing methods struggle with the high variability inherent in pituitary surgery durations.
Purpose of the Study:
- To develop and evaluate a novel spatio-temporal neural network model, PitRSDNet, for predicting Remaining Surgery Duration (RSD) in endoscopic pituitary surgery.
- To improve the accuracy and efficiency of RSD predictions by integrating workflow knowledge.
- To enhance patient care and minimize surgical theatre costs through better scheduling.
Main Methods:
- Development of PitRSDNet, a spatio-temporal neural network model leveraging historical surgical data.
- Integration of workflow knowledge through multi-task learning (predicting surgical steps and RSD concurrently).
- Incorporation of prior surgical steps as contextual information for temporal learning and inference.
Main Results:
- PitRSDNet demonstrated competitive performance improvements over traditional statistical and machine learning methods.
- The model was trained and evaluated on a new dataset comprising 88 endoscopic pituitary surgery videos.
- Findings indicate PitRSDNet enhances RSD prediction precision, particularly for outlier cases, by utilizing prior step information.
Conclusions:
- PitRSDNet offers a significant advancement in predicting Remaining Surgery Duration for endoscopic pituitary surgery.
- The model's ability to learn from workflow sequences and incorporate prior steps improves prediction accuracy and reliability.
- This approach holds potential for optimizing surgical scheduling, improving patient care, and reducing healthcare costs.
