Setup errors using daily CT and CBCT for imageguided stereotactic body radiation therapy SBRT

1
Setup errors using daily CT and CBCT for
image-guided stereotactic body radiation therapy
(SBRT)
Jianzhou Wu1, Martin Fuss1, and Warren
DSouza1,2 1Department of Radiation Medicine,
Oregon Health Science University, Portland,
OR 2Department of Radiation Oncology, University
of Maryland School of Medicine, Baltimore, MD
Purpose
Absolute alignment errors over 30 setups

• Immobilization devices, used for patients
  underwent stereotactic body radiation therapy
  (SBRT) improve patient alignment and
  repositioning accuracy
• The BodyFIX system (BodyFIX, Medical
  Intelligence, Schwabmuenchen, Germany) has been
  shown to perform favorably compared with other
  body immobilization systems
• In this study, we retrospectively compared
  patient alignments using daily simulation CT with
  patients immobilized within BodyFIX and
  alignments using on-board cone-beam CT (CBCT)

Materials and Methods
error shift from daily CT alignment shift
from CBCT alignment

• Ten patients who underwent SBRT of lung cancers
  were immobilized using BodyFIX system. For
  image-guidance, all patients were immobilized and
  scanned in the simulation suite. An
  image-guidance control CT was acquired on the
  dedicated simulation CT scanner (Philips
  Brilliance Big Bore 16, Philips Medical Systems,
  Cleveland, OH) before each fraction of treatment.
  Necessary positional setup shifts were
  established by comparison between original
  simulation CT and the respective control CT.
• Patients were then transferred onto a stretcher,
  and moved onto the linear accelerator treatment
  table within their BodyFIX for radiotherapy.
• Prior to treatment initiation, patients were
  scanned again using CBCT for alignment
  confirmation.
• This study investigated the capability of patient
  alignments using a daily simulation CT. Owing to
  the fact that patients were imaged in the
  simulation suite, and then moved to the treatment
  vault, we assigned the CBCT based alignment as
  the reference for accurate treatment setup.
• Alignment variations between modalities were
  evaluated by a two-step image registration, which
  registered both the image-guidance CT and the
  CBCT to the original simulation CT.
• First a rigid registration was performed to align
  the image-guidance CT (and CBCT) to the
  simulation CT. After rigid registration,
  patients were aligned with match in bony
  structures.
• Next a deformable registration was performed,
  which allows soft tissue deformation so that a
  better local match was achieve. After this step,
  patients were aligned with tumor-to-tumor match.
• After image registration, tumor contours
  delineated on the simulation CT were warped to
  the daily image-guidance CT (and CBCT) using the
  transformation fields saved from image
  registration.
• The center-of-mass (COM) of the mapped contours
  was calculated. The difference between the COM
  after deformable registration and that after
  rigid registration stands for the shift after
  soft tissue based alignment with respect to bony
  structure based alignment
• The setup optimization shifts derived from
  image-guidance CT based alignments were compared
  with the corresponding shifts from CBCT based
  alignments for the evaluation of setup accuracy.
  Particular interested was directed towards the
  question of detection of patient setup variations
  owing to transfer from the imaging system in the
  simulation suite to the treatment system in a
  different room.

Alignment deviations between modalities (30
setups in 10 patients)
A t-test was performed for alignments using
daily CT and CBCT over 30 setups. The
corresponding p-value was shown in the above table
Discussion

• Daily simulation CT based alignments were
  compared with CBCT image-guidance based
  alignments for patients immobilized using BodyFIX
  system and transferred between 2 rooms
• On average the alignment mismatch between the two
  modalities was less than 2 mm along the three
  principle axes
• A t-test demonstrated that alignments using the
  two methods were indistinguishable along the
  principle axes, except in the left-right
  direction
• This discrepancy may be related to one or two
  distinguishable observations. (1) The transfer of
  patient and immobilization system from CT table
  onto the stretcher and then again onto the linear
  accelerator table was always a lateral transfer.
  This particular form of transfer may be a cause
  the observed setup inconsistency.
• (2) Abrupt lateral motion of the linear
  accelerator treatment table related to moving the
  tabletop into a safe, collision free position for
  CBCT acquisition. The impact of the created
  momentum on stability of patient setup in an
  immobilization system is subject to ongoing
  testing.
• The averaged alignment discrepancy was well
  within the 5 mm margin, which was added to the
  ITV when creating the PTV to account for setup
  uncertainty.
• Therefore daily simulation CT based
  image-guidance for SBRT with patients
  immobilization and transfer using BodyFIX system
  provides for accurate and safe patient setup
  assessment.

Demonstration of evaluation process
Alignment using CBCT
Rigid registration
Deformable registration
Contours mapped after deformable registration
Contours mapped after rigid registration
Alignment using daily CT
Conclusion
Contour on simulation CT

• Patient alignments using daily simulation CT was
  consistent with the alignments using CBCT
• Alignment using image-guidance CT with patient
  immobilized using BodyFIX system provides a
  solution for image- guided radiation therapy on
  facilities without the on-board imaging capability

shift COM of mapped contour after deformable
registration COM of contour after rigid
registration