A Multiphase Technological Validation of a MRI Prostate Cancer Computer Autosegmentation Software Algorithm

Spencer Martin, Nikhilesh Patil, Stewart Gaede, Alexander Louie, Glenn Bauman, David D’Souza, Tracy Sexton, David Palma, Farzad Khalvati, Hamid Tizhoosh, George Rodrigues

Purpose:

Segmentation of the prostate gland on MRI imaging can be time consuming and may be subject to intra- and inter-physician reliability effects. The purpose of this investigation was to perform a rigorous technological assessment and validation of a software package/algorithm that can provide prostate autocontours on prostate MRI datasets.

Methods and Materials:

A three-phase validation strategy for the evaluation of a computer autosegmentation software algorithm (Segasist PMR-C, Segasist Technologies, Toronto ON) was employed....... *

Results:

..........*

Assessment of the pre- and post-autosegmented contours demonstrated a higher average DICE score of 0.94 suggesting that the algorithm was providing robust initial contours that required minimal editing..........*

Conclusions:

The Segasist PMR-C autocontouring algorithm has been validated to assist clinicians in the segmentation of prostate MRI datasets in terms of contouring time and accuracy. The software provides robust contours with minimal editing required by the user......... *

* Full text not available due to copy right restrictions