Pitfalls in Volume Rendering of Small Lesions for Intracranial Stereoetactic Radiosurgery

LIJUN MA1, Arjun Sahgal2, Alex Karokti3, Andrew Hwang4, Brian Wang5, Dennis C Shrieve5, David A Larson4

1UCSF 2University of Toronto 3University of Toronto, Toronto, Ontario, Canada 4University of California San Francisco, San Francisco, CA, USA 5University of Utah, Salt Lake City, UT, USA

Keywords: Dose distribution, cyberknife, gamma knife, radiosurgery, Brain Tumors

Interactive Manuscript

Ask Questions of this Manuscript:

   



Abstract

     
     
To determine device-dependent variability in volume rendering of small intracranial lesions via different intracranial stereotactic radiosurgery (SRS) modalities.
     
     Phantom and patient studies were carried out to determine the volumes of small objects rendered from different SRS systems that included Gamma Knife, Cyberknife, Novalis TX, Tomotherapy, Combo-linac systems such as Artiste/Trilogy/Synergy. For the phantom stuVolume rendering of small lesions based on two-dimensional Dicom structure set points showed significant variations among current SRS modalities. Such an uncertainty underscores the importance of benchmarking different SRS modalities for clinical trials involving dose-volume surrogates.dy, precisely machined acrylic spheres of varied sizes (0.64 cm to 3.81 cm in diameters) embedded in a styrofoam block were scanned per SRS imaging protocol on a CT scanner with varied slice thicknesses. These spherical objects were contoured on the Gamma Knife treatment planning system (LGP 9.0). The rendered volumes of these contours closely matched (within 0.2%) the physical volumes of individual objects. Such contours were then exported via Dicom RT data into all the intracranial treatment planning systems for volume rendering and comparison. Finally, such a procedure was repeated for a Gamma Knife SRS patient whose 12 intracranial lesions (0.4-1.2 cm in diameters) were exclusively defined by the senior clinician (DAL) of the study.
     For the phantom study, the mean discrepancy in the rendered volumes of the spheres against their actual physical volumes among all the SRS systems was 3.8%±2.6% (ranged from 0.1% to 20.8%). For the patient case, the mean discrepancy in the rendered volumes against the initial target volumes among studied SRS modalities was 7.1%±2.4% (ranged from -0.1% to 33.0%).
     
     Volume rendering of small lesions based on two-dimensional Dicom structure set points showed significant variations among current SRS modalities.  
     Such an uncertainty underscores the importance of benchmarking different SRS modalities for clinical trials involving dose-volume surrogates.


Acknowledgements

Project Roles:

L. MA (), A. Sahgal (), A. Karokti (), A. Hwang (), B. Wang (), D. Shrieve (), D. Larson ()