Related Experiment Video
Updated: Feb 22, 2026
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
43.7K
Finding discriminative and interpretable patterns in sequences of surgical activities
Germain Forestier1, François Petitjean2, Pavel Senin3
1MIPS EA 2332, University of Haute-Alsace, Mulhouse, France; Faculty of Information Technology, Monash University, Melbourne, Australia.
Artificial Intelligence in Medicine
|September 26, 2017
Summary
This study introduces a method to identify unique surgical practice patterns from surgery recordings. The approach accurately distinguishes between surgery locations, surgeon expertise levels, and even individual surgeons, enhancing surgical process understanding.
Area of Science:
- Medical Informatics
- Surgical Analysis
- Pattern Recognition
Background:
- Understanding surgical behaviors requires analyzing similarities and differences in surgical procedures.
- Surgery recordings, composed of low-level activity sequences, offer valuable data for surgical behavior analysis.
Purpose of the Study:
- To identify discriminative patterns within surgical practice recordings.
- To develop a method for analyzing surgical activity sequences to understand variations in surgical behaviors.
Main Methods:
- Utilized the vector space model (VSM) to represent surgical activity sequences.
- Split activity sequences into subsequences and computed relative frequencies using the tf*idf framework.
- Employed Cosine similarity for classifying sequences and identifying discriminative patterns.
Main Results:
- Successfully identified patterns discriminating surgery locations, surgeon expertise levels (expert vs. intermediate), and individual surgeons.
- Demonstrated the utility of tf*idf weight vectors for visualizing key patterns and highlighting significant surgical segments.
- Experiments were conducted on 40 anterior cervical discectomy (ACD) neurosurgeries.
Conclusions:
- The proposed method effectively identifies discriminative and interpretable patterns in surgical activity sequences.
- This approach aids in understanding surgical processes by automatically highlighting differences between groups of surgeons.
- Findings are crucial for improving surgical training and quality assessment.
