Automated Segmentation for Prostate Bed Contouring
Author Information
Author(s): Hwee Jeremiah, Louie Alexander V, Gaede Stewart, Bauman Glenn, D'Souza David, Sexton Tracy, Lock Michael, Ahmad Belal, Rodrigues George
Primary Institution: University of Western Ontario
Hypothesis
Can an automated atlas-based segmentation (AABS) engine reduce contouring time and inter-observer variability in prostate bed contouring?
Conclusion
The AABS tool results in time savings and shows good spatial overlap for certain organs, but requires improvement for the prostate bed and penile bulb.
Supporting Evidence
- The mean time for manual contours was 17.5 minutes, while for AABS it was 14.1 minutes.
- The AABS tool had a mean DSC of 0.67 for the prostate bed after editing.
- Inter-observer variability was significant with a p-value of 0.002 for the prostate bed.
Takeaway
This study shows that a computer program can help doctors outline areas for cancer treatment faster and more consistently, but it still needs some help for certain areas.
Methodology
Eighty post-prostatectomy patients were randomly selected, and contours were created using both manual and automated methods, followed by statistical analysis of the results.
Potential Biases
Attempts were made to minimize bias through blinding and the use of consensus guidelines.
Limitations
The study's findings may not be applicable to other practice groups due to the specific patient population used.
Participant Demographics
Eighty post-prostatectomy patients from the University of Western Ontario.
Statistical Information
P-Value
0.003
Confidence Interval
0.18-0.59
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website