Surgical Data Science Flashcards
evalotion of surgery pros and cons
+ Higher safety
+ Better patient outcomes
+ More illnesses are treatable
- Much higher complexity
- More training required
Surgical Data Science
Why
How
What
WHY:
* Improve the quality of interventional healthcare
HOW :
* Capture data (medical records, imaging, sensors)
* Processing, analysis and modelling of data
WHAT :
* Create smart systems that can predict events and clinical outcomes, assisting physicians
in decision-making, controlling devices, planning execution and prevention of errors
SDS challenges
- Data availability
- regulatory, technical, and
social factors (patient privacy and confidentiality)
- Missing standards to collect and share data efficiently
- For the annotation of data, we need medical experts, making it expensive - Heterogeneous data
- Robustness, reliability and interpretability
SDS – Applications
Decision Support
Context-aware assistance
Surgical Training
SDS – Potential Applications
+ Early warning when deviating from surgical plan
+ Prediction of adverse advents (e.g., bleeding)
+ Automatic surgical reporting and archiving
+ (semi-)automatic surgical robots
Challenges
Variability of patient anatomy and surgeon style
Limited available training data
Similarities between different phases
Modeling and Recognition in OR challenges
- Coverage despite occlusions
- Detected small objects
- Depth and Distance Perception
- Dynamic Range
improvements in Modeling and Recognition in OR
Multiple Cameras:
* Significantly reduces occlusions
* Details can be retrieved from the closest camera
RGB + Depth Cameras:
* Reduces geometric ambiguities
* Allows matching between points from different cameras
Multiple Exposures:
* Can capture information from both dark and bright parts
potential data sources
Room Multi RGB-D
Laparoscopy RGB
Tool data
Audiu cues
Patient data
Anasthesia
Surgical Data Science aim
to connect all the components to form an intelligent system
