Spiral CT Techniques, Volumetric Image Formation, and its Application to Colon Screening

1
Spiral CT Techniques, Volumetric Image Formation,
and its Application to Colon Screening

• Jennifer Dempsey

2
Introduction to Spiral/Helical CT

• Continuous acquisition of CT data
• Trajectory of X-ray beam traces out a spiral
• Single-slice available in 1989 multi-slice in
  1998

3
Detector Arrays

• Allow for increase in speed along z-axis
• Isotropic Arrays (GE, Toshiba)
• Use widths that are multiples of the detector row
  extent in the z-direction
• Work best for smaller collimations, but result in
  dead zones in the image with larger collimation
  widths
• Adaptive Arrays (Siemens, Marconi)
• Built anisotropically with fewer septa in the
  z-direction
• Higher geometric efficiency, but also increased
  complexity
• Allow for flexibility in slice widths

Isotropic (Fixed) Array
Adaptive Array
4
Adjustable Parameters

• Reconstruction Interval
• Very small reconstruction intervals give
  increased longitudinal resolution
• Tradeoff with processing time required and number
  of images to review
• If reconstruction interval is greater than the
  collimation, longitudinal resolution is equal to
  the reconstruction interval
• If reconstruction interval is less than the
  collimation, the longitudinal resolution improves
  up to a certain plateau
• Spiral CT allows for an increase in longitudinal
  resolution without increasing X-ray dosage, as
  required with conventional CT scanning techniques
• Collimation
• Spiral pitch

5
Spiral Pitch

• Single-slice spiral CT
• Ratio of table feed per rotation of the X-ray
  source to the collimated slice thickness
• Ranges from 0 to 2
• Less than 1 gaps in image
• Close to 2 blurring of image
• Multi-slice spiral CT
• Definition varies by manufacturer
• Includes number of simultaneously acquired
  slices (M)

6
Volumetric Image Generation

• Interpolate image data longitudinally to produce
  isotropic voxels
• Generate volumetric reformations
• Maximum intensity projections (MIPs)
• Shaded surface displays (SSDs)

7
Linear Interpolation Methods

• Single-slice spiral CT
• 360-degree linear interpolation method (360 LI)
• Uses data points that have been separated by a
  full 360-degree rotation of the X-ray tube
• Requires data from two full rotations to
  reconstruct an image slice
• Inadequate with respect to noise and artifact
  elimination
• Prominent blurring is visible along the
  longitudinal direction (z-axis)

8
Linear Interpolation Methods

• Single-slice spiral CT
• 180-degree linear interpolation method (180 LI)
• Interpolates projection data with opposite view
  angles
• Substantial increase in longitudinal resolution
• Possible to scan with a pitch greater than 11,
  which allows for more coverage within the same
  timeframe
• Increase in noise
• Followed by filtered backprojection

9
Linear Interpolation Methods

• Multi-slice spiral CT
• Same reconstruction principles apply - 180 MLI
• More common approach is to use z-filtering 180
  multi-slice filtered interpolation (180 MFI)

10
Volumetric Image Reformation

• Maximum intensity projections (MIPS)
• Map the maximum attenuation
  value along imaginary
  rays projected
  projected through the 3D matrix of
  interpolated
  image data to a gray
  scale image
• The projected rays can be oriented in
  any anatomic plane (coronal,
  sagittal, transaxial)
  
• Shaded surface displays (SSDs)
• Generated on the basis of some intensity
  threshold
• If a voxel in the interpolated 3D matrix has an
  attenuation value greater than the threshold, it
  is set to white otherwise, it is set to black.
• Using depth perception given by shading
  techniques, the thresholded data is rendered into
  a 3D image.

11
Cartesian Coordinate Interpolation

• Convert all projection rays into parallel beam
  format
• A correction term is applied since the projection
  rays are tilted relative to the x-y plane
• Compute reconstructed image in Cartesian
  coordinates based on these rays
• Perform this for entire volume to create a 3D
  matrix of data points in Cartesian coordinates

12
Introduction to CT Colonography/Virtual
Colonoscopy

• Alternative to conventional colonoscopy
• For proper screening for polyps
• Collimation values less than 5 mm
• Helical pitch less than 2
• Reconstruction intervals between 1 and 2

13
Virtual Colonoscopy

• Benefits
• Higher patient tolerance, acceptance
• Reduced patient risk
• Faster examination time
• Can view entire colon
• Can also view regions of the colon from any
  direction
• Ability to detect extracolonic findings
• Challenges
• Necessity of prone and supine scanning
• Differentiation between benign/malignant
• Potential for follow-up colonoscopy

14
Feasibility of Virtual Colonoscopy for Colon
Screening

• An extensive study (Pickhardt, et al, 2003) on
  1233 adults to compare the performance of each
  exam type
• Relied on three-dimensional reconstructed images
• A similar study performed by Cotton, et al (2004)
  compared conventional and virtual colonoscopy
  results for 600 patients
• Wide variation in results between facilities
• Relied primarily on two-dimensional reconstructed
  images

15
Sensitivity Results
16
Fly-Through Algorithm

• Region Growing
• Select seed voxel within colon, examine its
  26-connected neighborhood region add to region
  if voxel meets intensity and threshold criteria
• May require multiple seed points if collapsed
  regions these regions are then connected with
  26-connected cubic splines
• Bubble Removal
• Identify 6-connected regions of unidentified
  voxels with 6-connectivity to the surface of the
  segmented region
• Fill in these regions with 6-connectivity region
  growing

17
Fly-Through Algorithm

• Initial Path Determination
• Manual selection of start and end points
• Optimal path searching algorithm used to find
  shallowest descent (will not cut corners)
• Thinning
• Successively remove surface voxels until only the
  central axis remains
• Smoothing/Resampling
• Convolution of x,y, and z coordinates to smooth
  the path
• Camera Orientation
• Point camera to where the path first disappears
  around a corner

18
Conclusions

• Spiral CT Techniques have
• Improved longitudinal resolution
• Decreased scanning time
• Numerous interpolation and reconstruction
  techniques have been developed for volumetric
  image generation
• Using conversions to parallel beam format, spiral
  CT can be represented in Cartesian coordinates
• Spiral CT is a feasible tool for use in colon
  screening in conjunction with virtual colonoscopy

19
Questions?