Clinical Implementation of a LowDose 4D CT Acquisition Protocol

1
Clinical Implementation of a Low-Dose 4D CT
Acquisition Protocol

• L Xing, T Li, E Schreibmann, B Thorndyke, G
  Tillman, A Boyer, A Koong, K Goodman

Department of Radiation Oncology Stanford
University
2
Abstract
Four-dimensional (4D) CT scans, acquired
synchronously with a respiratory signal, provide
not only the 3D spatial information, but also
temporal changes of the anatomy as a function of
the respiratory phase during the imaging, and can
therefore be employed in 4D treatment planning to
explicitly account for the respiratory motion.
However, it usually delivers 1015 times more
radiation dose to the patient as compared to the
standard 3D CT, since multiple scans at each
couch position are required to obtain the
temporal information. In this work we propose a
statistic method to obtain high quality 4D CT
with low tube current, hence reducing the
radiation exposure of patients.
3
High Radiation Dose

• Temporal information comes from the repeated
  scans of patients.
• In a 4D acquisition, each slice is scanned for a
  complete respiration cycle, the radiation
  exposure is 1020 times higher than 3D CT.
• Patient dose in conventional 3D CT 2 cGy for
  the chest or abdomen. In 4D acquisition, it could
  be up to 40 cGy in one chest exam.

4
Lower 4D CT Radiation Dose

• We develop a novel method to enable the 4D scan
  to be performed at lower x-ray tube current
  without much loss of the image quality, hence
  reducing the patient radiation dose.

5
Major issue lower current causing more
statistical noise
100 mA 4D scan
10 mA 4D scan
Phase 0
Phase 20
Phase 40
Phase 60
Phase 80
6
Solution

• If the object is stationary, each phase is then
  an independent repeated scan. Stacking them
  together by simple average will result in an
  improved image that is equivalent to a high-mA
  scan CT image.
• (Ideally, image by averaging five 10-mA data is
  equivalent to a 50-mA scan)
• When motion exists in the object, registration is
  needed to bring different phases to the same time
  point before stacking can be taken.

7
Deformable Registration

• Checkerboard display of two superimposed images
  of the inhale and exhale phases

before registration
after registration
8
Registration Errors

• Ideally, if the images are registered perfectly
  to one phase, then each of them can be considered
  a repeated scan for this phase. Therefore, it is
  natural to average them to get an improved image.
  In reality, simple averaging will often lead to
  blurring in the image, due to errors in
  registration, resulting in suboptimal solution.

9
Registration Errors

• Ideally, if the images are registered perfectly
  to one phase, then each of them can be considered
  a repeated scan for this phase. Therefore, it is
  natural to average them to get an improved image.
  In reality, simple averaging will often lead to
  blurring in the image, due to errors in
  registration, resulting in suboptimal solution.

10
Statistic Stacking

• We developed a statistic method to incorporate
  the available information into a penalized
  weighted least square (PWLS) objective function
  to achieve the optimal estimation of the true
  image at each phase (4D-PWLS method).

11
PWLS Solution

• Iterative algorithm

The weights of temporal neighbors are
proportional to the normalized cross-correlation
12
Low-Dose 4D CT Results Phantom (1)
10 mA after processing
10 mA
100 mA
Phase 0
Phase 20
Phase 40
Phase 60
Phase 80
13
Low-Dose 4D CT Results Phantom (2)
10 mA after processing
10 mA
window width / level 200/80
window width/ level 1500/150
14
(No Transcript)
15
Low-Dose 4D CT Results Patient

• End-inspiration phase
• SNR increased 2.22 times.

Original
Processed
16
Low-Dose 4D CT Results Patient

• End-expiration phase
• SNR increased 2.07 times.

Original
Processed
17
Conclusions

• A novel technique to reduce the radiation dose in
  4D CT has been developed.
• Improvements in image quality have been observed
  in both phantom and patient studies.
• The proposed method is being applied to 4D PET
  imaging and 4D treatment planning.

18
Acknowledgement This research is supported in
part by Varian Medical Systems, Department of
Defense (DAMD17-03-1-0023) and the National
Cancer Institute (1R01 CA98523-01).