ASTRO poster IBLS

1
Performance Analysis of anOptoelectronic
Localization System forMonitoring Brain
Lesioning with Proton BeamsFadi Shihadeh(1),
Reinhard Schulte(2), Keith E. Schubert(1), Pani
Chakrapani(3)(1)California State University,
San Bernardino, CA(2)Loma Linda University
Medical Center, Loma Linda, CA(3)University of
RedlandsThis research was funded by the Henry L.
Guenther Foundation

• Introduction
• In preparation for a clinical functional
  radiosurgery program with highly focused proton
  beams, we have designed a versatile system for
  anatomical lesioning that can be used both in
  clinical and research applications (Fig. 1 and
  2).
• Here we describe a performance study of an
  optoelectronic system that serves to facilitate
  the alignment of the anatomical target to the
  proton beam axis.

Fig. 2. CT isodose plan of a multi-field proton
radiosurgery treatment for lesioning of the
pituitary gland of a rat.
Fig. 1. Rat immobilized in holder for functional
radiosurgery procedures.

• Optoelectronic Positioning and Alignment Control
  System (OPACS)
• The OPACS optically tracks 2 retro-reflective
  marker sets, attached to the target and to the
  proton beam axis, respectively (Fig. 3).
  Alignment software calculates the distance of the
  target point from the beam axis and provides
  output to the positioning system to correct any
  offset.
• The opto-electronic localization system (OLS),
  which is an integral part of the OPACS, consists
  of 3 high resolution cameras and tracking
  software (Vicon 260, Vicon Motion, Systems, Ltd,
  Oxford, UK).
• Before each measurement session, the Vicon system
  is calibrated by waving a wand of 2 reference
  markers of known distance through the measurement
  space.

Fig. 4. Camera configuration scheme. The three
cameras formed an equilateral triangle and were
oriented so that their central rays met at
iso-center under a common angle f. The two
configurations tested were f 50 and f 90 .
Fig. 3. Optical localization system for
functional radiosurgery procedures.

• Performance Study
• 15 retro-reflective markers were repeatedly
  tracked measured distances were compared to
  those obtained by a certified dimensional
  measurement lab (DML) and marker shifts were
  measured after prescribed shifts with
  XYZ-microstages .
• For each session, we repeatedly measured the
  distance of each marker from the center of
  gravity (CG) of all other markers (distance
  error) and the measured marker displacement after
  prescribed shifts (shift error).
• Experimental parameters included 2 different
  camera configurations (Fig. 4) and 4 wand-waving
  (calibration) techniques.

• Results
• During initial runs, a small random scaling
  factor between 0.98 and 1.02 was found by
  comparing measured and DML distances. Henceforth,
  we determined the scaling factor for each
  measurement trial and used it to correct the CG
  distance accordingly.
• Mean CG distance errors were of the order of 0.1
  mm and independent of camera configuration and
  the calibration technique. Standard deviations of
  the marker means ranged from 0.16 mm to 0.21 mm
  (Table I).
• The between-marker variation of the distance
  error was the largest source of variation (Fig. 5
  top), followed by the inter-session variation.
  The use of different camera calibration
  techniques did not significantly increase the
  inter-session variability.
• Mean shift errors were lt 0.01 mm and their
  standard deviations were between 0.014 mm and
  0.022 mm (Table II). There was no dependence on
  camera configuration or calibration technique.
• Practically all of the variance in the shift
  error could be contributed to intra-session
  (trial) variation (Fig. 5 bottom).

Fig. 5. Decomposition of the standard deviation
(SD) of the distance error (top) and shift error
(bottom) into components due to marker, session,
and trial (intra-session) variations.

• Conclusion
• The Vicon optical localization system provides
  sub-millimeter accuracy and precision in the
  localization of retro-reflective marker systems.
• The precision of the localization of a given
  marker relative to the average position of other
  markers depends mostly on the marker quality
  itself and to a lesser degree on variation
  between measurement sessions.
• Relative shift measurements with respect to a
  prior position are inherently more precise than
  distance measurements as they do not depend on
  the marker quality or inter-session variability.