Tomographic Acquisitions :Tomographic Acquisitions Presentation by: Claudy Gilles Gabriel Marrero Dailiuber Mendez


Objectives :Objectives List SPECT acquisition mode. Compare circular and body-contour orbits. Describe reorientation planes used in cardiac SPECT. Discuss volume rendering techniques and their advantages for display presentation. Describe surface rendering of SPECT data sets. Describe the general techniques used in automatic reorientation. Discuss gray scale versus color display. Describe techniques used to acquired gated cardiac SPECT studies.


Introduction :Introduction SPECT images are created by mathematically combining many planar scintigrams (projections) that have been collected at many angles around the body. Theory of SPECT requires that the scintillation camera acquire enough images of the patient at different angles.


SPECT Data Acquisition Modes :SPECT Data Acquisition Modes Most SPECT systems move in a fashion called step-and-shoot as they orbit the patient. Camera moves, stops, acquires an image, moves again, stops and acquires another image. Optimum number of stops is determined by the system resolution. Dead time – time during which the camera is moving and turned off


Continuous Acquisition :Continuous Acquisition Continuous orbit camera acquire counts at all times even while the camera moves. Continuous motion cameras move slowly and smoothly while counts from designated arcs are collected to form the planar projections. The smaller the arc used for each projection , the less the blurring from the camera’s motion


Continuous Step-and-Shoot :Continuous Step-and-Shoot Combines step-and-shoot with continuous acquisition Camera is stopped at discrete projection angles but is not turned off during rotation Minimizes the blur seen with continuous-rotation acquisition but maximizes the number of collected counts in a given time


Long AcquisitionTime to complete a SPECT study using step-and-shoot and continuous acquisitions :Long AcquisitionTime to complete a SPECT study using step-and-shoot and continuous acquisitions


Short AcquisitionTime to complete a SPECT study using step-and-shoot and continuous acquisitions :Short AcquisitionTime to complete a SPECT study using step-and-shoot and continuous acquisitions


180-Degree vs. 360-Degree Data Acquisition :180-Degree vs. 360-Degree Data Acquisition 360-Degree orbits are preferred except for cardiac SPECT Heart is located forward and to one side resulting in great deal of attenuation when camera is behind patient Reconstructions from 180-degree acquisitions have higher resolution and contrast than those from 360-degree acquisitions 180-degree acquisitions are not truly complete so occasional artifacts are seen that can be avoided with 360-degree acquisitions


Circular vs. Body-Contour Orbits 1 :Circular vs. Body-Contour Orbits 1 Circular orbits gamma camera is limited to rotating in a circle Position of detector is fixed at a given radius with respect to the center of rotation (COR) Forcing the camera to stay at a fixed radius at one angle may result in it being very far from the body at another angle


Circular vs. Body-Contour Orbits 2 :Circular vs. Body-Contour Orbits 2 Body-contour orbits Keep the camera as close as possible to the body throughout the tomographic acquisition, maximizing resolution Radius is changed angle by angle Simplest body-contour orbits assume the shape of an ellipse


Circular vs. Body-Contour Orbits 3 :Circular vs. Body-Contour Orbits 3 For general SPECT, the results of body-contour orbits are superior to those of circular orbits Use of body-contour orbits for cardiac SPECT is not as widely accepted Resolution varies from angle to angle so resulting reconstruction may be different from one created with a circular orbit


Other Factors :Other Factors Three important variables Size of the image pixels Average number of counts collected for each pixel Number of views obtained When a zoom factor is used, the field of view is reduced and the pixels of the image are used more efficiently


Size of the image pixels :Size of the image pixels Pixel size should be less than 1/3 system resolution A 128 x 128 pixel matrix is preferred for most low-energy studies because it gives a pixel size of about 4 mm Pixels 6 to 6.5 mm in size have proved to be adequate for cardiac SPECT


Average number of counts :Average number of counts Number of counts in each pixel is related to system resolution Higher the resolution , the lower the number of counts obtained per pixel Number of counts should be maximized without exceeding radiation dose limits or the patient’s ability to remain immobile


Number of views :Number of views Number of views depends on the resolution and the pixel size If too few views are taken, resolution is lost and artifacts may be introduced


SPECT Image Display :SPECT Image Display


Color vs. Gray Scale :Color vs. Gray Scale Different color scales may be used to highlight different features in SPECT images Certain color scales may help distinguish normal from abnormal Gray scale is better for highlighting very dim objects SPECT images should always be scaled and displayed in a reproducible and standardized manner to ensure reproducibility in qualitative analysis


Image Reorientation: transaxial images :Image Reorientation: transaxial images Transverse or Transaxial slices Natural products of rotational tomography are images that represent cross-sectional slices of the body perpendicular to the imaging table


Image Reorientation: Longitudinal Images :Image Reorientation: Longitudinal Images SPECT images can be reconstructed so that the image slices are contiguous and the slice thickness is equal to the pixel size Pixels are called voxels once recognized that they have depth and width Images may then be thought of as a 3D matrix of information about the patient


Image Reorientation: Oblique Images :Image Reorientation: Oblique Images Computer may be used to extract images at any orientation and is not limited to x,y, and z dimensions This is called oblique images Examples include short- and long-axis images used in myocardial tomography May also be used to retrospectively obtain a desired image orientation for brain imaging


Three-Dimensional Displays :Three-Dimensional Displays Most common display mode is a 2D interactive slicing program that allows a user to view a 3D data set one slice at a time Since this program is limited to determining an accurate idea of size or shape, efforts are underway to generate true 3D displays These displays fall into two categories Volume Rendering Surface Rendering


Volume Rendering :Volume Rendering Advantage of visualizing an object in 3D without having to explicitly identify the surface Most common type is Maximum Intensity Projection (MIP) Developed by Wallis and Miller Reconstructed transaxial slices are stacked to form a 3D tomographic volume, MIP rotates this volume into desired viewing angle and extracts the maximum pixels onto a 2D image plane Useful in blood pool and tumor imaging procedures


Surface Rendering :Surface Rendering Refers to methods that display an organ or region based on explicitly detected boundaries Requires segmentation Separates the organ from the background or nearby structure Surface-rendered models are generally displayed using standard computer graphics packages, which allows rotation and translation When this technique is applied with appropriate boundary detection algorithims, it is the most realistic way to view results of quantitation


Cardiac SPECT Quantifications :Cardiac SPECT Quantifications


Cardiac Reorientation :Cardiac Reorientation Most cardiac images are viewed in a standard format consisting of short-axis, horizontal long-axis, and vertical long-axis slices Short-axis slices are also necessary for some automatic perfusion quantification algorithms Generation of these standard sections from the original transaxial images has been performed interactively and requires the user to mark the location of the left ventricular axis


Vertical Long-axis slices :Vertical Long-axis slices Using a display of a transaxial slice through the middle of the left ventricle, the position of the long axis is denoted with a line drawn by the user The 3D set of transaxial sections is resliced parallel to the long axis and perpendicular to the transaxial slices Resulting oblique image is called a vertical long-axis slice


Horizontal Long-axis slices :Horizontal Long-axis slices Using a display of a midventricular vertical long-axis image, the user denotes the left ventricular long axis The 3D block of vertical long-axis slices is recut parallel to the denoted long-axis and perpendicular to the stack The resulting oblique cuts are called horizontal long-axis slices Contain the left ventricle with its base toward the bottom and apex toward the top


Short-axis slices :Short-axis slices Slices perpendicular to the denoted long axis and perpendicular to the vertical long-axis slices are cut from the stack Result are short-axis slices Contain left ventricle with its anterior wall toward the top Serial short-axis slices are displayed from apex to base, left to right


Automatic Reorientation 1 :Automatic Reorientation 1 Two approaches start by identifying the left ventricular region in the transaxial images, using a threshold-based approach that includes knowledge of the expected position, size, and shape of the left ventricle (LV) Once region has been isolated, approach described by German et al uses the original data to refine the estimate of the myocardial surfaces


Automatic reorientation 2 :Automatic reorientation 2 Best-fit gaussian function is used to estimate the myocardial center for each profile After further refinement, the resulting midmyocardial points are fitted to an ellipsoid , the long axis of which is used as the final LV long axis


Automatic reorientation 3 :Automatic reorientation 3 Mullick and Ezquerra use a more complex heuristic technique After this has been accomplished, they use the segmented data directly to determine the long axis Binary image is tesselated into triangular plates and the normal of each plate is used to point to the LV long axis


Automatic reorientation 4 :Automatic reorientation 4 Slomka et al register the original image data to a template image in which the orientation of the LV is known and standardized Template is created by averaging a large number of registered normal patient data sets, and separate templates are created for males and females


Perfusion Quantification :Perfusion Quantification All commercially available quantification methods for myocardial perfusion in SPECT are based on the idea that sampled counts in the myocardium of a patient’s image can be compared with similar counts sampled from a set of normal objects Circumferential profile – graph containing maximum count values against the angle at which they were encountered By comparing a patient’s circumferential profile angle by angle and slice by slice to the normal limits, perfusion defects can be pointed out automatically


Perfusion Quantification 2 :Perfusion Quantification 2 Analysis is performed and normal limits are created for the stress study and for a normalized difference between the stress and rest studies Separate normal limits must be created for males and females Abnormal areas or defects seen in the images that persist on the rest study are considered fixed Few studies have compared these various approaches to perfusion quantifications


Cardiac Display :Cardiac Display


Polar Maps :Polar Maps Polar maps – bull’s eye displays, another way to view circumferential profiles Give a quick, comprehensive overview of the circumferential samples from all slices by combining them into a color-coded image Most apical slice forms center of the polar map Most basal slice of the left ventricle makes up the outermost ring of the polar map Allows for immediate and comprehensive viewing of the quantitative results of the entire myocardium


Polar Maps 2 :Polar Maps 2 Use of color can aid the identification of abnormal areas at a glance Although they offer a comprehensive view , polar maps distort the size and shape of the myocardium and any defects Numerous improvements in basic polar map display have helped to overcome some of these problems


Three-Dimensional Cardiac Displays :Three-Dimensional Cardiac Displays Can be used to overlay results of perfusion quantification onto a representation of a specific patient’s left ventricle In the most basic form, pixel locations of the maximum-count myocardial points sampled during quantitation are used to estimate the myocardial surface Such displays can be rotated in real time and viewed from any angle


Three-Dimensional Cardiac Displays 2 :Three-Dimensional Cardiac Displays 2 The biggest disadvantage of 3D displays is that they require more computer screen space than polar maps Example, the entire left ventricle can be visualized in a single circular polar map but only one side of the left ventricle can be seen when it is displayed using 3D graphics


References :References Christian, P. E. (2007). nuclear medicine and PET/CT. In k. m. waterstram-rich. salt lake city: mosby elsevier. http://www.impactscan.org/rsna2004.htm http://mp-spect.com/examples.htm


questions :questions True or false 1. SPECT systems only move in a fashion called step-and-shoot as they orbit the patient.


Slide 43:False Most SPECT systems move in a fashion called step-and-shoot as they orbit the patient.


Slide 44:True or false 2. Continuous orbit camera acquire counts at all times even while the camera moves.


Slide 45:True


Slide 46:True or false 3. There is less lost time in Short AcquisitionSPECT study using step-and-shoot and continuous acquisitions.


Slide 47:False In a Short Acquisition SPECT study using step-and-shoot and continuous acquisitions, there is more lost time.


Slide 48:Multiple choice 4. When a zoom factor is used, the field of view is _______ and the pixels of the image are used more _______ used, than ever Increased, sparingly Reduced, efficiently Blurred, laboriously


Slide 49:c. reduced, efficiently When a zoom factor is used, the field of view is reduced and the pixels of the image are used more efficiently


Slide 50:Multiple choice 5. Which pixel matrix is preferred for most low-energy studies because it gives a pixel size of about 4 mm 64 x 64 256 x 256 128 x 128 512 x 512


Slide 51:c. 128 x 128 A 128 x 128 pixel matrix is preferred for most low-energy studies because it gives a pixel size of about 4 mm


Slide 52:True or false 6. Surface rendering refers to methods that display an organ or region based on explicitly detected boundaries


Slide 53:true


Slide 54:True or false 7. Most cardiac images are viewed in a standard format consisting of short-axis, medial long-axis, and vertical long-axis slices


Slide 55:False! Most cardiac images are viewed in a standard format consisting of short-axis, horizontal long-axis, and vertical long-axis slices


Slide 56:True or false 8. Normal limits must be created equally for males and females when performing a stress study


Slide 57:False Separate normal limits must be created for males and females for the stress study and for a normalized difference between the stress and rest studies


Slide 58:True or false 9. Polar maps allows for immediate and comprehensive viewing of the quantitative results of the entire myocardium


Slide 59:true


Slide 60:True or false 10. The biggest advantage of 3D displays is that they require less computer screen space than polar maps


Slide 61:False The biggest disadvantage of 3D displays is that they require more computer screen space than polar maps