Clinical evaluation of an MRI-to-ultrasound deformable image registration algorithm for prostate brachytherapy

Amani Shaaer; Melanie Davidson; Mark Semple; Alexandru Nicolae; Lucas Castro Mendez; Hans Chung; Andrew Loblaw; Chia-Lin Tseng; Gerard Morton; Ananth Ravi

doi:10.1016/j.brachy.2018.08.006

Clinical evaluation of an MRI-to-ultrasound deformable image registration algorithm for prostate brachytherapy

Brachytherapy. 2019 Jan-Feb;18(1):95-102. doi: 10.1016/j.brachy.2018.08.006. Epub 2018 Oct 2.

Authors

Affiliations

¹ Department of Physics, Ryerson University, Toronto, Ontario, Canada.
² Department of Medical Physics, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada; Department of Radiation Oncology, University of Toronto, Toronto, Ontario, Canada.
³ Department of Medical Physics, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada.
⁴ Department of Radiation Oncology, University of Toronto, Toronto, Ontario, Canada; Department of Radiation Oncology, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada.
⁵ Department of Medical Physics, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada; Department of Radiation Oncology, University of Toronto, Toronto, Ontario, Canada. Electronic address: ananth.ravi@sunnybrook.ca.

PMID: 30287271
DOI: 10.1016/j.brachy.2018.08.006

Abstract

Purpose: Identifying dominant intraprostatic lesions (DILs) on transrectal ultrasound (TRUS) images during prostate high-dose-rate brachytherapy (HDR-BT) treatment planning is challenging. Multiparametric MRI (mpMRI) is the tool of choice for DIL identification; however, the geometry of the prostate on mpMRI and on the TRUS may differ significantly, requiring image registration. This study evaluates the efficacy of an in-house software for MRI-to-TRUS DIL registration (MR2US) and compares its results to rigid and B-Spline deformable registration.

Methods and materials: Ten patients with intermediate-risk prostate cancer, each with mpMRI and TRUS data sets, were included in this study. Five radiation oncologists (ROs) with expertise in TRUS-based HDR-BT were asked to cognitively contour the DIL onto the TRUS image using mpMRI as reference. The contours were analyzed for concordance using simultaneous truth and performance level estimation algorithm. Similarity indices, DIL volumes, and distance between centroid positions were measured to compare the consensus contours against the contours from ROs and the automated algorithms; registration time between all contouring methods was recorded.

Results: MR2US registration had the highest dice coefficients among all patients with a mean of 0.80 ± 0.13 in comparison to rigid (0.65 ± 0.20) and B-Spline (0.51 ± 0.30). The distance between centroid positions between simultaneous truth and performance level estimation contour and MR2US, rigid, and B-Spline contours were 5 ± 2, 7 ± 5, and 18 ± 11 mm, respectively. The average registration time was significantly shorter for MR2US (11 ± 2 s) and rigid algorithm (7 ± 1 s) compared to ROs (227 ± 27 s) and B-Spline (199 ± 38 s).

Conclusions: The efficacy of integrating an MRI-delineated DIL into a TRUS-based BT workflow has been validated in this study. The MR2US software is fast and accurate enough to be used for DIL identification in prostate HDR-BT.

Keywords: HDR brachytherapy; Image registration; MRI; Prostate cancer; TRUS.

Publication types

Research Support, Non-U.S. Gov't
Review

MeSH terms

Algorithms*
Brachytherapy / methods*
Humans
Magnetic Resonance Imaging
Male
Prostatic Neoplasms / diagnostic imaging
Prostatic Neoplasms / radiotherapy*
Radiotherapy Dosage
Radiotherapy Planning, Computer-Assisted
Radiotherapy, Image-Guided / methods*
Software
Ultrasonography
Workflow