SINGLE-PHOTON EMISSION

COMPUTED TOMOGRAPHY (SPECT)

 INTRODUCTION

 Rotating gamma camera acquire multiple planar views of the radioactivity in

an organ.
 SPECT cameras rotate about the patient in order to determine the three-
dimensional distribution of radiotracer in the patient.
 The data are then processed mathematically to create cross sectional views of
the organ.
 SPECT utilizes the single photons emitted by gamma-emitting radionuclides
such as 99mTc, 67Ga, 111In, and 123 I.
Most common SPECT system consists of a typical gamma
camera with 1 to 3 NaI (Tl) detector heads mounted on a gantry , an
online computer for acquisition and processing of data and a Display
system.
 SPECT CAMERAS:

 First, the SPECT camera is constructed so that the head can

rotate either stepwise or continuously about the patient to
acquire multiple views.

 Second, it is equipped with a computer that integrates the

multiple images to produce the cross-sectional views of the
organ.
 Advanced SPECT camera designs have more than one head or are constructed with a
ring of detectors.

 In the case of the single and multiple head cameras, the heads are mechanically rotated
around the patient to obtain the multiple projection views.

 Ring detectors have a ring of individual small crystals or a single ,donut-shaped crystal
that does not rotate.

 SPECT images are reconstructed on a matrix of 64×64 or 128×128 pixels.

 ANGLE OF ROTATION OF HEADS

Single-headed cameras:

 Must rotate a full 360◦ to obtain all necessary views of most organs.

Two-Headed Cameras :
 Each head of a double headed camera need to rotate only half as far ,180 ◦ .

Three-Headed cameras:
 Triple-headed camera only rotates 120◦ to obtain the same views.
Three headed SPECT camera
 SPECT
Process

 The essence of SPECT is to obtain 2D images called projections ,from

different angles; to reconstruct from those images the object activity
distribution into a 3D image matrix

 Then displays 2D slice of the 3D image matrix

 ACQUISITIO
 Na gamma camera that can rotate in a circle
The first requirement for SPECT is

 The projections are most commonly acquired in a step-and-shoot format or continuous

 Step-and-shoot is the standard method of tomographic projection views.

 In this technique, each projection view is acquired at each angular stop(position).

 There is a short pause of a few seconds between views to allow for the automatic rotation of the camera head
to the next stop.

 Total acquisition time-1408s.

 In this continuous, data is collected over one or several sequential 360 ◦ rotations.

 There are no pauses; rotation is continuous.

 The camera rotated a full 360̊ every 140.8s.

 Ten such rotation provides 1408s of imaging time.

CONTINUOUS ACQUISITION
 Circular orbits -
 Most acquisition performed.
 The camera head is rotated at a fixed distance from the center of
CIRCULAR the body.
VS
ELLIPTICAL ORBITS
( For improving sensitivity) Elliptical orbits -
 Allow the camera head to more closely follow the contour of the
body and therefore stay closer to the organ being imaged.( better
resolution)
CIRCULAR ORBITS

ELLIPTICAL ORBITS
  2 different approaches have been used

RECONSTRUCTION  Analytic Reconstruction – filtered

backprojection
 Iterative Reconstruction
 ANALYTIC
RECONSTRUCTION
 With gamma camera at an angle , each location on the camera face sees a column of
tissue

 This column of tissue has some amount of activity in each small subsections

 The total amount of radioactivity in column is called ray-sum

 The value’s registered by the gamma camera is therefore related to the ray sum

 A single 2D projection then consist of a number of pixels , each of which estimate a ray
sum value
  A simplistic way to go about reconstruction is to take values
in each pixel and apply it to all of the points along the
equivalent column in the reconstructed 3D image matrix

 A single hot spot with in a circular object is imaged in four

Unfiltered projections
Backprojection
 In each projection hot spot is seen as increased activity

 Which is presented as a peck in each 4 count profiles

 The count profiles are backprojected to image matrix

 The absent knowledge of the location of the hot spot , the pixel value from the count
profile is instead put in to every voxel in the column perpendicular to the projection
plane

 We will call this Unfiltered Backprojection (UBP)

 UBP produce an image in which the count value at a particular location is the total of
the ray-sum
 UBP has two drawbacks’

I. UBP puts count in the 3D image matrix where there was no

activity in the corresponding area of the object

II. Area of elevated count density is larger than the area of

increased activity in the object and its size increase as we add
more projection

 Blurring makes the constructed image difficult to interpret

  UBP produce a blurred image with counts in
incorrect position

Filtered  This is because UBP uses the low frequency

data repeatedly as it backproject the 2D
Backprojection projection

 To correct this we use a filter called the RAMP

FILTER
 Ramp filter multiplies the image data at each frequency

 Low frequency data are suppressed and high frequency

image data are amplified

 Ramp filter is sufficient to remove the blurring that

produce UBP and also restore the sharp edges of the 3D

 Ramp filter is only applied in the transaxial plane because

that is the only plane that experiences blurring
 If our image had no noise we could stop with the ramp filter

 But NM images always have noise

 This noise is more prominent in the high frequency part of

the image

 Application of ramp filter not only amplifies the sharp edges

and small objects visible but also amplifies image noise in an
unacceptable level
 It is necessary to cut off the linear ramp around the point where the
patient data disappear into the noise

 For this reason ramp filter is combined with a second filter that rolls
off the ramp filter

 Roll off filters;

I. Low-pass filter
II. Band -pass filter
III. High- pass filter
 By removing the noise the low pass filter performs a smoothing
function

 A number of low-pass filter have been applied to SPECT

 They are usually known by the name of developer

 Eg; Hann and Hamming filter

  The desire to more accurately correct SPECT
images has pushed the development of a
different family if reconstruction technique
called iterative method

ITERATIVE  The essence of iterative method is to make a

series of estimate of the radiopharmaceutical
RECONSTRUCTI distribution in the object
ON
 Each estimate in turn is modified ,based on
comparison to the acquired projections to
produce the next estimate

 The process over many iterations approaches

the best match to the acquisition data
  It starts with an initial guess as to the
radiopharmaceutical distribution ; assuming a
uniform distribution

Iterative  The computer then creates estimated 2D

Cycle projections using the initial guess, additional
information related to the physics of
imaging ,and patient specific attenuation map
 This is the first estimated 3D matrix
 This estimated projection are then compared to the measured 2D
projections obtained in the study

 The discrepancies between the two are used to modify the estimated
3D image matrix

 This modified image matrix is then used as the starting point for the
next iteration

 The process is repeated until the estimated projections are very

similar to the measured projections

 Each trip through this cycle counts as one iteration

  The final step in SPECT study is display of the
3D image matrix

IMAGE  Display devices are almost entirely 2D

DISPLAY
 Because our information is in 3D we want to
display in the way that convey the 3D spatial
relationship of the structure
  Once the 3D image matrix has been reconstructed the
tomographic slices are immediately available.

 We can view axial, coronal and sagittal planes

2D Display
Technique  We can display each plane as a series of 2D slice of the 3D
image matrix

 2D slice represent the highest resolution version of data

 In NM identification of an abnormality is considered only when the
abnormality is seen in at least two planes

 Dynamic triangulation allows the reader to select a point in one

slice of the 3D image matrix and click on it with cursor

 The computer then displays the 2 slice containing same voxel

Dynamic triangulation is exceptionally helpful in
SPECT because the limited amount of anatomic
information in the image makes all 3 planes
necessary for image interpretation
  Several techniques are used to
3D Display show visually realistic
Technique representations of the 3D image
matrix in a 2D format
 SURFACE RENDERING generate a solid model of exterior pixels
hiding the interior pixel from the view

 First step in surface rendering is to identify the surface of the organs

by segmenting the image into either object or background

 Surface generated is highlighted according to its reflection

capabilities , as it were a solid object illuminated by external light

 Surface rendering is helpful when the organ being images is most

easily seen as a continuous surface
 VOLUME RENDERING reprojects the 3D image matrix to create
2D projections

 Can be viewed in a rotating cine display

 Projection is accomplished by a ray-tracing technique

 Rays are traced at different angles and different locations

 All the ray tracings at a particular angle can be combined to form

a reprojection of 3D matrix image
 The process result in images that are quite similar to the raw SPECT
projections

 But the image have lower noise and greater contrast

 In contrast to surface rendering image , volume rendering images are

semitransparent

 It allow the viewer to see one structure through another

 The resulting image when displayed to rotating cine format gives the
impression of organ volume and depth
 MIP is very similar to volume rendered image

 MIP image use only the hottest voxel along each ray to
generate the projections

 Projections are created and displayed in rotating cine fashion

 MIP images are very helpful for hot-spot imaging because the
area of increased activity can be correlated to their correct 3d
location
CLINICAL APPLICATIONS OF SPECT

Cardiac (myocardial perfusion)-

For CAD, and Heart muscle damage
Brain (cerebral perfusion)-
Cerebrovascular disease , Dementia , seizure
disorder , Brain tumors , psychiatric disease.

ONCOLOGY

Imaging of infection and inflammation

Measurement of liver and kidney function

THANK
YOU