Semiautomatic segmentation for prostate brachytherapy: dosimetric evaluation.

Area of Science:

• Medical Imaging
• Radiation Oncology
• Computational Anatomy

Background:

• Manual prostate segmentation in transrectal ultrasound (TRUS) images is crucial for accurate radiation therapy planning.
• Manual segmentation is time-consuming and subject to inter-observer variability, potentially impacting treatment consistency and efficacy.

Purpose of the Study:

• To evaluate the feasibility of replacing manual prostate segmentation with a semiautomatic approach using a tapered ellipsoid model in TRUS images.
• To assess the volumetric and dosimetric accuracy of the semiautomatic segmentation compared to manual segmentation.

Main Methods:

• A tapered ellipsoid model was employed for semiautomatic segmentation of the prostate in TRUS images.
• Volumetric evaluation was performed by comparing semiautomatic contours with physician-reviewed manual contours.
• Dosimetric assessment involved comparing treatment plans generated using semiautomatic contours with those based on manual contours.

Main Results:

• The semiautomatic method achieved an average volume error of less than 7% (1-Dice similarity coefficient) across 140 cases.
• Dosimetric analysis showed a difference of less than 5% in V(100) parameters between plans based on semiautomatic and manual contours.
• The average total segmentation time per case was reduced to 2-4 minutes, including algorithm initialization and contour refinement.

Conclusions:

• The proposed semiautomatic segmentation algorithm is a viable alternative to manual segmentation for prostate cancer radiation therapy planning.
• The method offers significant benefits, including reduced segmentation time, improved consistency, and decreased reliance on user expertise.
• The algorithm ensures smooth, symmetric contours, contributing to both volumetric and dosimetric accuracy.