-
Multi-slice CTPrinciples and Perspectives Mindy M. Horrow, MD,
FACR Director of Body Imaging Albert Einstein Medical CenterAll
photos retain the copyrights of their original owners 2005 Mindy
Horrow, MDTo advance slide, use spacebar; use left arrow for
previous slide; use browsers back button to exit program
-
Before CTEntire areas of body inaccessible to radiography
(brain, retroperitoneum, etc.) Some useful diagnostic procedures
were either potentially harmful or considerably uncomfortable
(exploratory laparotomy, pneumoencephalography)
-
Principles of CTRadiographic tube emits x-rays while rotating
axially around patient Array of detectors on opposite side of
patient detects x-rays transmitted through patient Computer
algorithms use digitized data from detectors to create axial
tomographic images of body. CT = tomography + algorithms + high
speed digital computers
-
Tomography- CT actually eliminates unwanted material, outside of
scan plane instead of just blurring it (1921-conventional
tomography) Reconstruction algorithms- Fast Fourier Transformation:
allows mapping of function of space into a function of frequency
using the theorem that any function can be decoded into a sum of
sine and cosine functions (described by Fourier in 1811)Principles
of CT
-
Development of Workable CT Scanner1963- Cormack in S. Africa
develops algorithm for accurate reconstruction of images from
radiographic projections 1971- Hounsfield, a computer engineer in
England produces first working CT scanner used clinically on
patients. Produced 70 head CTs in 6 mos, at 4 min per slice,
recorded on magnetic tape with two days reconstruction time per
case. Cormack and Hounsfield awarded Nobel Prize in medicine and
physiology in 1979
-
Sir Godfrey Hounsfield with a prototype CT scanner in 1974
-
Head CT circa 1975 with 128 x 128 matrix
-
Radiology: Volume 119, 1976Davis, Taveras, New, et al. Diagnosis
of Epidermoid Tumor by Computed Tomography Hahn, et al The Normal
Range and position of the Pineal Gland on Computed Tomography
Huckman, Ramsey, et al. Computed Tomography in the Diagnosis of
Pseudotumor Cerebri Messina, Potts, et al. Computed Tomography:
Evaluation of the Posterior Third Ventricle OConnor, et al.
Computed Tomography in a Community Hospital Sagel, Stanley, Evens.
Early Clinical Experience with Motionless Whole Body Computed
Tomography
-
Sagel, Stanley, Levitt, et al. Computed Tomography of the Kidney
Radiology 124:359-370, 1977Computed tomography is an extremely
accurate method of obtaining more definitive diagnostic information
about a renal mass discovered on a urogram. Benign renal cysts are
readily distinguished from solid renal neoplasms, and CT is often
valuable in characterizing possible juxtarenal masses. The cause of
a nonfunctioning kidney(s) on a urogram can often be discerned, and
hydronephrosis is easily detected.
-
Proliferation of CTBy 1976, 3 years after Hounsfields
publications, 22 companies were manufacturing CT scanners By 1979
1000 scanners were operating in 50 countries Competition produced
rapid technological sophistication Introduction of fan
beam-scanning decreased scan time from 300 sec to 2 sec per slice
in 4 years
-
Conventional CT scannersEmploy fan of x-ray beams and a large
detector array 3 types of gantries: translate-rotate,
rotate-rotate, rotate-stationary Involves alternating patient
translation and x-ray exposure Each rotation of x-ray tube
generates data from which a corresponding axial image is
reconstructed
-
Helical (spiral) CTSimultaneous patient translation and x-ray
scanning generates volume of data X-ray beam traces a helix of raw
data from which axial images must be generated Each rotation
generates data specific to an angled plane of section To create
true axial image, data points above and below desired section must
be interpolated to estimate value in axial plane Thus, interval
between reconstructed transsexual images can be chosen
retrospectively
-
Technological considerations of helical CTSlip-ring technology
(no electrical cables connecting gantry to ground) allows source
detector assembly to rotate continuously. Previously, frequent,
abrupt changes between scans were necessary to permit winding and
unwinding of cables More robust x-ray tubes and generators were
developed to allow high tube current for prolonged duration. Also
needed to be lightweight enough to be mounted in slip ring
gantry
-
Comparison of single slice and multi-slice CTDetector
configuration Reconstructions Detector design Definition of pitch
Pitch and image quality Spatial resolution
-
Configuration of detectorsSS- long, narrow array with length of
single detector aligned in z axis MS- detector array segmented in z
axis, a mosaic Allows for simultaneous acquisition of multiple
images in scan plane with ONE rotation
-
Mosaic Detector 16 cells in Z direction --each cell 1.25 mm (in
Z) 16 cells (Z) x 912 cells (transverse) = 14592 total cells Signal
collected from 4 channels/2 flex connectors
-
ReconstructionsSS- reconstruct images of SAME thickness with
different image indexing (table increment intervals) MS- acquire 3D
raw data that are contiguous in space. Therefore can reconstruct
images at various thicknesses AND at different intervals If image
index < image thickness, results in overlapping slices Must have
raw data available for any type of reconstruction
-
Multi-slice detector design (GE 4 slice scanner)16 equal
elements in z axis, 20mm maximum collimator width * Can acquire 1,
2, or 4 images per rotation For example: with collimator at 10mm
can make 4 images @ 2.5mm, 2 images @ 5mm or 1 image @ 10mm
Thinnest slice thickness that can be reconstructed depends entirely
on combination of slice thickness and table speed* A single 1.25
detector is made of two .63 detectors
-
Axial Configurations4 x 2.5 mm4 signals collected from eight
1.25 mm detectors with 2 detectors contributing to each signal 2.5
mm is the minimum slice thickness because two 1.25 mm detectors are
combined per signal Cells can be combined to form 4 slices @ 2.5 mm
each or 2 slices @ 5 mm each or 1 slice @ 10 mm12341214i mode =
each set of 2 cells becomes a slice @ 2.5 mm each1i mode = 4 sets
(of 2 cells each) are combined to form 1 slice @ 10 mm2i mode = 2
sets (of 2 cells each) are combined to form 2 slices @ 5 mm
each
-
Pitch SS = table travel per rotation image thickness If table
travel > slice thickness, pitch > 1 MS = table travel per
rotation total active detector width* * = x-ray beam
collimation
-
Table travel/ rotation = 7.5mm Image thickness = 5mm Pitch = 7.5
= 1.5 5.0 Table travel/ rotation = 7.5mm Four Images with thickness
= 2.5mm Pitch = 7.5 = .75 (4 x 2.5) SS MS
-
GE Definition of PitchTable travel per rotation = 7.5 = 3 single
image slice thickness 2.5 (High quality mode) i.e.. When 4 images
are acquired per tube rotation, associated table travel is 3 times
image width
-
GE Definition of PitchTable travel per rotation = 15 = 6 single
image slice thickness 2.5 (High speed mode) i.e.. When 4 images are
acquired per tube rotation, associated table travel is 6 times
image width
-
Pitch and Image QualitySS- Image quality decreases as pitch
increases MS- GE scanners have unique property of forming images
with particularly good quality at 2 specific pitch values, HQ and
HS
-
Pitch and NoiseTo reconstruct image, projections must be
collected over 180 gantry rotation and fan angle of x-ray beam (45
), about 2/3 of spiral Since reconstruction algorithms need fixed
number of projections to make image and since pitch only affects
how these projections are distributed in spiral, not the number of
projections, pitch does not affect noise No difference between SS
and MS
-
Image Quality: Contrast ResolutionAbility of imaging system to
detect a single structure that varies only slightly from its
surroundings Related to noise AND pitch Less noise, fewer
distractions, increased ability to perceive low density object
Contrast resolution in x-y plane as pitch for SSCT but does NOT
change for MSCT
-
Contrast ResolutionSS- pitch causes broadening of
slice-sensitivity profile. Scanner needs to have enough projections
to reconstruct slice and is forced to seek them outside of specific
z axis. Some projections may not pass through object in question
and results in under-sampling which blurs object
-
Contrast ResolutionMS- pitch does not broaden SSP because at
least one of multiple rows of detectors passes into x-y plane
containing object in question. Because of multiple detectors,
highly unlikely that projections distant from imaging plane will be
needed.
-
Z axis resolution DoseSS- increased pitch decreases z axis
resolution MS- increased pitch has little effect on z axis
resolutionSS- increased pitch decreases radiation dose MS increased
pitch, machine compensates with increased mA and dose does not
change
- Dose-Pitch RelationshipFor SS, if pitch>1, dose decreases
pitch1, dose similar pitch
-
Reconstruction AlgorithmsSS and MS are similar First step done
by machine, z axis interpolator works on raw data to weight
projections nearest the slice location most heavily Second step
selected by user: for soft tissue images, want to suppress noise
and increase low contrast sensitivity. For bone want higher
contrast.
-
Protocols for MSCTImage thickness, detector configuration,
collimation, table speed, interval, reconstruction algorithms IV
contrast parameters Length of acquisition Technique: kV, mA, sec
Reconstruction algorithm FOV- for scan, for display
-
Problems/Pitfalls in Protocol DesignTiming of bolus and data
acquisition Preset filming Pseudo-enhancement Venous artifacts
Increasing numbers of tiny lesions
-
TimingRoutine chest protocol with 20 cm coverage, table speed
11.25 mm/rotation, takes 14 sec for entire scan at 0.8 sec/slice
Using 40 sec prep delay If injection rate is 2cc/sec, use 108 cc If
injection rate is 2.5, use 128 cc
-
Filming Because of high levels of vascular enhancement, classic
soft tissue windows will not be appropriate for all organs. Lesions
may be obscured in organs that enhance brightly such as kidney and
arteries (pulmonary emboli, dissection flaps)
-
Other issuesPseudo-enhancement of renal cysts surrounded by
parenchyma becomes a greater problem because of higher levels of
renal enhancement Increasing numbers of tiny lesions in lung,
liver, etc. Are these metastases? Venous artifacts, which simulate
thrombus become more obvious and frequent
-
Can I reconstruct thinner slices than those printed on image?PE
protocol 2.50mm/7.50 1.5:1 Image 2.5mm thickness Table speed = 7.5
mm/rotation Pitch = 1.5:1 (HS mode) 7.5 1.5 = 5mm collimation With
4 slices per rotation, detector size must be 1.25mm and therefore
this is thinnest slice thickness that one can reconstruct
-
16 slice scannerRoutinely 360 rotation in 0.5 sec (798 data
points) Can go to 0.4 sec rotation for cardiac scanning For larger
patients, increase rotation time Using the large-large FOV, each
pixel is 1mm in x-y plane. Thus each vowel is 1 x1 x1mm = ISOTROPIC
SCANNING Can also achieve isotropic scanning with small FOV (head,
neck, extremity) in which each voxel is 0.5mm
-
16 slice scannerHelical pitch = table distance per rotation /
slice thickness 15mm / 1mm = 15 Beam pitch = helical pitch / image
thickness 15 / 16 = 0.938
-
16 slice scannerProspective Gating: 0.4 sec gantry speed.
Machine counts 5 cycles, calculates R-R interval, takes 0.25 sec
scan, ending prior to beginning of next R wave. Requires heart rate
< 80 bpm. Table moves during next R-R interval. Each scan covers
4 images @ 3mm thickness. Retrospective gating = gated
reconstruction. Helical acquisition with ECG over raw data. When
ECG is at a given point, take data at that time to make image.
-
MSCTExamples of Unique Protocols
-
LungAbnormal CXR- survey exam with 5mm sections, but choose
detector set of 4 1.25mm so that retrospective thin slices through
a small nodule can be obtained without re-imaging Airway disease-
single breath hold with 1 or 1.25mm collimation. Evaluate trachea
with overlapping 3-5mm sections, use 1mm sections to assess small
airways. Combine inspiratory & expiratory views for physiologic
evaluation, air trapping PE- 1.25- 2.5mm, HS mode scan average
thorax in 10-12 sec. Use bolus tracking. Helpful to view as
reconstructions.
-
AbdomenPorta hepatis- because of complicated anatomy &
oblique orientation, 3mm sections recon w 50% overlap curved or
coronal reformats Liver and pancreas- multiple phases in single
breath holds Kidneys- 3D reconstructions similar to IVU and
angiography
-
Musculoskeletal System0.5mm slice thickness can result in
isotropic data set such that x-, y- and z-axes are equal in size
(Will be standard on 8 and 16 slice scanners) Trauma- if scanning
chest, abdomen, pelvis can change FOV, recon to thinner slices,
change to bone algorithm and do 2D and 3D recons to review spine,
without re-imaging as spine protocol
-
Head and NeckElimination of direct coronal imaging Reformats can
avoid artifacts from teeth 3D displays for trauma and congenital
anomalies 3D reformats can provide endoscopic views of larynx,
hypopharynx and to calculate tumor volumes Angiographic and
perfusion studies
-
Post Processing ApplicationsHuge numbers of images can be
generated from original data set and reformatted in different
planes, surface displays, angiographic techniques, virtual
endoscopy Issues of how to view and store images
-
QUIZ
-
1. For SSCT with image thickness of 2.5 mm and table speed of
4.0 mm/rotation, the pitch = __________ 2. For MSCT with tech
requesting 4 slices with 3.75 mm thickness and table speed of 11.25
mm/rotation, pitch = ________________
-
a. SSCTb. MSCTc. bothd. neither3. Re-indexing, or creating
overlapping slices is possible on 4. Reconstruction images to
different slice thicknesses is possible on 5. X-Y axis resolution
(image quality) does not change significantly with higher pitch
on
-
a. SSCTb. MSCTc. bothd. neither6. Noise increases with
increasing pitch on 7. As pitch increases, > 1, the radiation
dose to the patient decreases on 8. The reconstruction algorithm
involves 2 steps, the z axis interpolator and application of
specific algorithms such as smooth, standard, bone, etc on
-
1. For SSCT with image thickness of 2.5 mm and table speed of
4.0 mm/rotation, the pitch = 1.6 2. For MSCT with tech requesting 4
slices with 3.75 mm thickness and table speed of 11.25 mm/rotation,
pitch = 0.75
-
a. SSCTb. MSCTc. bothd. neither3. Re-indexing, or creating
overlapping slices is possible on c 4. Reconstruction images to
different slice thicknesses is possible on b 5. X-Y axis resolution
(image quality) does not change significantly with higher pitch on
b
-
a. SSCTb. MSCTc. bothd. neither6. Noise increases with
increasing pitch on d 7. As pitch increases, > 1, the radiation
dose to the patient decreases on a 8. The reconstruction algorithm
involves 2 steps, the z axis interpolator and application of
specific algorithms such as smooth, standard, bone, etc on c
-
ReferencesBrink JA, Heiken JP, et al. Helical CT: Principles and
Technical Considerations. Radiographics 1994; 14:887-893 Friedland
GW and Thurber BD. The Birth of CT. AJR; 167: 1365-1370 Silverman
PM. multi-slice Computed Tomography- A Practical Approach to
Clinical Protocols. Lippincott Williams & Wilkins, 2002;
Chapter 1: 1-29 Thanks to GE for providing images
-
The End