ImageGuided Alignment Verification with Submillimeter Precision for Functional Proton Radiosurgery

1
Image-Guided Alignment Verification with
Submillimeter Precision forFunctional Proton
Radiosurgery

• Mahesh Neupane (1), Richard P. Levy(2) , Michael
  F. Moyers (2), Keith Schubert (1), Fadi Shihadeh
  (1) , and R.W. Schulte (2)
• (1) California State University, San Bernardino,
  CA
• (2) Loma Linda University Medical Center, Loma
  Linda, CA

2
Introduction

• High-energy proton beams (250 MeV), due to their
  exquisitely sharp lateral penumbra with minimal
  widening as the beam penetrates the patient, are
  suitable for functional radiosurgery procedures
  such as pallidotomy.

3
Purpose

• Guiding a narrow proton beam with a proton gantry
  that weighs about 90 tons to a target with
  submillimeter precision requires a sophisticated
  alignment verification system.
• We have developed a camera and marker-based
  system for this purpose and report on its initial
  performance in phantom tests.

4
Proton Gantry Patient Positioner

• Isocentric gantry, 35 feet tall,90 tons
• 100 MeV 250 MeV protons
• 2 mm isocentric accuracy
• 6-DOF robotic patient positioner

5
System Overview

• Three high-resolution CCD cameras (Vicon Motion
  Systems, Inc., Oxford, UK) in an equilateral
  triangular configuration were focused on a
  patient-based marker system (caddy), attached to
  the stereotactic halo, and a proton-based marker
  system (cross) attached to the proton beam
  delivery cone.

6
System Components
Marker Caddy
Cone Marker Cross
7
System Components
Camera System
Phantom Base
8
Experimental Method
Align Simulated target with known stereotatic
coordinates to Simulated beam
Measure residual offset
Calculate beam axis and transform
to stereotactic system
Calculate coordinate transformation
Capture cone and caddy markers
Calculate distance between beam axis
and simulated marker
Correct for residual offset to calculate system
error
9
Experimental Setup

• The markers, consisting of 5-mm ceramic balls
  covered with retro reflective tape, were the only
  structures visible to infrared-light strobed
  cameras.
• Based on the marker coordinates in the
  camera-based coordinate system, coordinate
  transformation software was written that computes
  the position of the proton beam axis in the
  stereotactic system.

Markers
10
Phantom Base and Coordinate Transformations

• Two different types of coordinate transforms were
  used a geometry-based orthogonal transform and a
  least-squares based transform.
• The patient was simulated by a phantom base
  providing a (non-reflective) ceramic marker
  target in 15 discrete positions in stereotactic
  space.

Phantom Base
11
System Error

• For each performance experiment, the target
  marker was aligned to the center of a 1-cm laser
  beam emerging from the proton delivery cone to
  better than 0.2 mm.
• The residual offset between the beam axis and the
  target measured by the camera system was
  interpreted as the system error.

12
Results

• The camera system was able to reproduce
  laboratory-inspected distances between the
  markers of the two marker systems with a mean
  error /- standard deviation (SD) of -0.23 /-
  0.33 mm for the caddy and 0.00 /- 0.09 mm for
  the cross.
• The alignment system error was sensitive to the
  type of coordinate transformation used
• In a series of 15 independent tests (5 target
  positions, 3 tests per position), a
  non-orthogonal least squares transform resulted
  in an unacceptable mean error (/- SD) of 25 /-
  8 mm (range, 14 mm - 36 mm).
• On the other hand, an orthogonal transform
  preserving geometric distances was associated
  with a much smaller error of 0.6 /- 0.3 mm (0.2
  mm - 1.6 mm).

13
System Error
Orthogonal transformation
Least-squares based transformation
14
Conclusion

• An image-guided alignment verification system has
  been tested under clinical conditions.
• The system error approaches but has not yet
  reached the goal limit of less than 0.5 mm.
• Various optimization options, e.g., different
  marker arrangements, will be applied to achieve
  this goal.

15
Acknowledgment

• This research was funded by the
• Henry L. Guenther
• Foundation