Encrypted login | home

Program Information

Automated Interstitial Brachytherapy Catheter Localization From Volumetric MR Data

no image available
G Pernelle

G Pernelle*, T Kapur, A Mehrtash, T Penzkofer, A Damato, L Barber, E Schmidt, A Viswanathan, R Cormack, Brigham and Women's Hospital, Boston, MA

SU-C-103-5 Sunday 1:00PM - 1:55PM Room: 103

Purpose: This work seeks to validate an automatic method for extracting interstitial catheter locations from volumetric MR data.

Methods: A 3D b-SSFP (balanced Steady State Free Precession) MRI sequence was used to enhance catheter detection, utilizing a controlled enlargement of size caused by magnetic susceptibility artifacts . Starting with a user-provided needle-tip location, the needle extraction algorithm searched the MR image for segments that maximized the "needle likelihood" and iteratively fit Bezier curves to these segments. To validate the geometry of extracted catheters, a phantom was constructed from transparent gel wax. An obturator was placed at the center of a transparent plastic container, and wax gel was melted and poured around it. A Syed-Neblett template was then placed at one end of the obturator and 12 MR compatible catheters were inserted into the gel. The catheter trajectories extracted from the MR scan by the algorithm were compared to a CT scan. The algorithm was applied to patient scans under an IRB approved protocol.

Results: The volumetric b-SSFP sequence produced an artifact enhancement that enabled brachytherapy catheters to be distinguished from other signal voids. Comparison in a phantom of catheter locations extracted automatically from MR to those extracted manually from CT showed agreement of 3.5 mm Hausdorff Distance. When applied to patient scans, the presence of low signal regions between template and the target tumor required some manual intervention to ensure proper catheter extraction.

Conclusion: An automatic algorithm for extracting interstitial catheter locations from MR scans has been developed. Using a b-SSFP sequence allowed an iterative curve fitting algorithm to track MR artifacts from user provided tips to template. This algorithm should prove helpful to facilitate MR based treatment planning for interstitial brachytherapy.


Funding Support, Disclosures, and Conflict of Interest: NIH P41 EB015898

Contact Email: