Nuclear cardiac imaging is widely employed in patients with coronary artery disease (CAD) or those at risk for this illness. This is somewhat remarkable since nearly all nuclear cardiology imaging is currently performed on instrumentation which is designed for everything from whole body to brain scans. This results in high cost for an instrument which is not optimized for the cardiac application, and in some cases is actually unable to perform certain types of nuclear cardiac studies. The high cost often makes it prohibitive for an individual cardiology practice to have its own instrument, leading many cardiologists to rely solely on echo cardiography. Thus we believe the need exits for a tomographic nuclear camera that is optimized for imaging the heart and is much less expensive than the currently available systems.

       The typical Single Photon Emission Computer Tomographic (SPECT) system used today is a general purpose device with one to three sodium-iodide [NaI(Tl)] detectors which rotate sequentially, with great precision, through about thirty angular positions. Most of the design characteristics of these SPECT systems have been selected to facilitate whole body imaging. A small, moving object, such as the heart, is not well suited for imaging by a device optimized to perform whole body imaging. Much of the detector surface area is wasted in that each image shows considerable uptake in structures far beyond the heart itself. The slow process of step-and-shoot imaging makes cardiac imaging a very prolonged process. Since the set of images is not acquired simultaneously, performing first pass studies or attempting to quantitate the rate of tracer uptake is not possible. Gated studies can also be difficult to obtain reliably and reproducibly by this sequential modality.

       Our specific goal is to build and demonstrate the clinical utility of a stationary or NonRotational Single Photon Emission Computed Tomographic (NRSPECT) system which is designed specifically for cardiac imaging. Such a system should be able to produce superior studies of the heart at significantly reduced cost in comparison with standard SPECT systems. The functional characteristics of our proposed NRSPECT system as optimized for cardiac imaging heart include:

1) Simultaneous acquisition of all views done through an array of pinhole collimators. This has several advantages over sequentially acquired views. First, it allows dynamic flow studies (e.g. first pass studies) to be done tomographically. Second, gated studies can be performed more reliably. Third, simultaneously acquiring the data shortens the acquisition time to about half that of a SPECT study (10-12 minutes for a static perfusion study and 20 minutes for a gated study). Finally, the use of pinhole collimation itself has the advantage of being geometrically perfect and very inexpensive to build.


2) Improved image statistics and full energy spectrum retention. Each individual image (10-12 minute acquisition) contains about 1 million counts, or about an order of magnitude more than in cardiac images acquired by SPECT, with each pixel retaining 128 channels of energy information. Static, frame-mode images are acquired as (128 x 128 x 128) integer arrays representing (X,Y and Energy). List-mode acquisition stores four 8-bit channels (X, Y, E and EKG) for each event. The improved statistical quality of these images, along with the additional energy information, facilitates and improves scatter and crosstalk correction. This allows simultaneous imaging of Tc-99m Sestamibi and Tl-201 (simultaneous dual isotope myocardial perfusion studies), and reduces the need (and time expended) to perform attenuation correction.


3) Electrical outputs of the detectors are directly interfaced to the computer(s). All the command and control functions, and signal processing are performed in software. This keeps the cost of the system as low as possible, allowing design modifications and upgrades to be accomplished quickly and inexpensively. For cost effectiveness we plan to use readily available low cost/high performance NaI(Tl) detector modules.

       In summary, the NRSPECT system we propose is intended to meet the nuclear imaging requirements for studies performed within the cardiologist’s office. Most cardiac practices operate a treadmill stress testing facility as part of their practice. Such facilities are frequently augmented with nuclear and/or echocardiographic imaging which are used both as screening tools and to monitor results of treatment for CAD. Many practices, however choose to use only stress echo because of its lower cost and familiarity with this imaging technology. Therefore, we plan on building our system at a cost which is competitive with stress echo and designed for ease of use with the flexibility necessary to support a high throughput volume of high quality nuclear studies.

       Each year, according to the American Heart Association, about 450,000 people in the U.S. die of causes related to CAD. While improvements in medical management and interventional treatment of CAD have made progress in reducing this number, the need for earlier diagnosis and better monitoring of the results of treatment continues. Stress echo studies are felt by some to fill this need for inexpensive diagnostic screening tool in application to CAD. Echo has an up-front economic advantage and is therefore commonly used in the cardiologist’s office. However, while both are roughly comparable in clinical accuracy, the echo cardiographic diagnosis of coronary artery disease requires the visualization of segmental left ventricular dysfunction. For that group of patients with a suitable echo “window”, segmental dysfunction can be directly viewed as a consequence of altered myocardial blood flow with stress. Nuclear imaging, however, is able to directly appreciate altered blood flow distribution itself and should therefore be more sensitive and specific. Bringing down the cost of nuclear studies will increase the attractiveness and availability of nuclear imaging to cardiologists. Nuclear studies also have other diagnostic and technical advantages over echo cardiography (Verani, 1994).

       Nuclear cardiac imaging is a well established technique capable of measuring the relative distribution of blood flow to the myocardium as well as providing information about the viability of the myocardial cells. The current standard of practice for nuclear cardiac imaging involves the use of dual or triple-headed systems with parallel hole or fan beam collimation. The requirement for rotation of the heads has two disadvantages, the data is acquired sequentially rather than simultaneously and the center of rotation is difficult to precisely control. The first disadvantage leads to certain types of studies not being possible or difficult to acquire, while the second leads to high system costs and potential reconstruction artifacts. Also, the collimator induced artifacts and detector inefficiencies lead to other problems when imaging a small, moving object like the heart.

       We believe that the NRSPECT system we are proposing will have the following benefits over the conventional rotating, parallel hole SPECT system: 1) The ability to acquire dynamic flow studies, which are otherwise unattainable; 2) An improved ability to do gated studies (corrected for respirational motion); and 3) The ability to do simultaneous Tc-99m and Tl-201 perfusion studies; and 4) Significantly reduce the cost associated with nuclear cardiology.

1) Dynamic flow studies cannot be acquired on current SPECT systems tomographically. When acquiring data sequentially, the underlying assumption is that the object (the heart) is stationary and unchanging. This is not strictly true. The heart is beating, there is upward creep as the level of breathing changes, the patient himself may be moving and the distribution of the radioactive tracer as well as the position of the heart may be changing with time. These can lead to artifacts in the final image (Friedman et al., 1989; Botvinick et al., 1993; Parker, 1993). While these studies are currently done by planar techniques, overlying or underlying activity can corrupt the data. Background activity can cause errors in the ejection fraction (EF) of up to 25% for the left ventricle and up to 5% for the right ventricle (Gal et al., 1986; Holman, 1988). With simultaneously acquired images, it is possible to study the uptake phase of tracer concentration in the myocardium in three dimensions. First-pass dynamic studies also provide better quantitation of ejection fractions, assessment of valvular patency and evaluation of cardiac shunts in the heart.

2) Gated studies of static distributions of both blood-pool tracers and radioactive metabolites located in the myocardium are often difficult to perform with current SPECT due to technical limitations (mostly due to statistics) and/or clinical complications (arrhythmias and limited patient tolerance). Patient movement (Botvinick et al., 1993) and upward creep (Friedman et al.., 1989) have been shown to be a cause of artifacts. Many motion correction methods have been proposed, but none to date have gained widespread acceptance. It should prove easier to correct motion in a system where all views are seeing the same motion (simultaneous acquired data) rather than where each view may see the heart in a different location in space and time (as in sequentially acquired data). Shorter acquisition times should also help reduce patient motion. Arrhythmias also cause problems since the beat-to-beat morphology is changing with time. Data acquired simultaneously has the advantage that each view sees the heart under the same conditions. The NRSPECT system acquires the data in list-mode giving the system the ability to do a posteriori beat selection. This can even allow different studies to be made from the same data set. For example, a patient with many premature ventricular contractions (PVC’s) can have a set of images made of only the normal beats and a separate set of images made from only the PVC beats (Koss et al., 1997b). This approach may give electro-physiological information concerning the origin of the contraction and the conductional pathway.

3) A simultaneous dual isotope protocol wherein the patient is under the camera for only a single sitting can be done by first injecting Tc-99m for the rest image, followed by stress injection of Tl-201 and acquiring the two images simultaneously. This has several obvious benefits (see editorial by DePuey, 1993). First, it eliminates errors in repositioning and setting of the acquisition parameters. Second, it cuts down the patient imaging time, allowing greater throughput of patients and is more convenient and comfortable for the patient. Third, heart motion (both body movement as well as translation and rotation of the heart) is identical for the stress and rest images. Another advantage of this protocol is that the Sestamibi image represents a true resting image. Thus the accuracy of this method should be much greater than current protocols. Unfortunately, the cross-talk contamination by scattered photons from Tc-99m into the Tl-201 energy window results in a significant reduction of the severity of Tl-201 defects (Kiat et al., 1994). While several commercial vendors are interested in simultaneous dual isotope capability, none have placed a product on the market largely due to the problem of crosstalk between the two isotopes. In all of the currently practiced sequential dual isotope protocols, the patient must be positioned for imaging twice which creates potential for error and artifacts (DePuey, 1993), and is less convenient for the patient and for the practice.

       The simultaneous dual isotope imaging can be successful only IF the entire energy spectrum is retained AND the images have sufficient counts (Little et al., 1996). Having only a few fixed energy windows, as is done on current SPECT systems does not provide sufficient data to perform adequate crosstalk correction between the Tc-99m and Tl-201 photo peaks (Moore et al., 1996; DePuey, 1993). We have successfully used this protocol on a single detector 7-pinhole system and extended the protocol to the multiple-detector pinhole system under Phase I.

4) Cost reduction will be accomplished in several ways. For instance, eliminating the need to move the detectors and performing all command and control functions, and signal processing in software significantly reduces the cost. Off-the-shelf AD boards and Pentium class PCs are inexpensive and are easy to upgrade. Pinhole collimators are easier and cheaper to build and maintain than parallel or fan beam collimators. Parallel hole collimators require very high precision, which makes them difficult to build and greatly increases their cost. Gantet et al., (1997) showed that very small defects in a parallel hole collimator can lead to significant uniformity artifacts. Defective collimators have been shipped (Yashizumi et al., 1990) and are often not tested for defects with the same diligence as camera detectors. The septa are also easily damaged, creating additional defects with use. Pinhole collimators are less expensive to build and do not have a significant problem with physical defects. Finally, the simultaneous dual isotope approach allows greater patient throughput, resulting in lower operating costs.

       Operating costs will also be lowered by enabling the user to update correction schemes rather than requiring a service call. We use a uniformity correction method which employs spatial shift vectors (Johnson et al., 1996), instead of schemes based on division by flood images (skimmers). This method simultaneously corrects both linearity distortions and spatial nonuniformities based only on the data from a set of three full-field flood images taken for isotopic energies over the energy range of interest. Updating the corrections is easily done by the user instead of requiring a service call. It does NOT alter the inherent sensitivity of the imaging system as is done with divisional schemes. This method of calibration is also an important prerequisite for providing identical imaging characteristics for Tc-99m and Tl-201.

       The Technology Marketing Group (TMG,1998) has recently published the results of a questionnaire from 5,101 hospital and non-hospital sites. This data presents the following picture of the market potential for our device. It states that 16 % of all nuclear medicine imaging procedures are performed outside of the hospital (2.1 million out of a total of 12.9 million procedures). Out of these 2.1 million non-hospital procedures, 71% or 1.5 million cardiac procedures are performed outside of the hospital (1.35 million are perfusion studies and 150,000 are other cardiovascular procedures). There are currently about 700 dedicated cardiac imaging labs in this country. The TMG survey also pointed out that out of the 10,820 installed cameras, only 3% of the cameras have triple-head SPECT capability, 21% are dual-head SPECT, 55% are single head SPECT and 21% are planar only. About two thirds of the dual headed systems (about 1500 units) can position the two detectors orthogonally for cardiac SPECT imaging. Demonstrating the feasibility of performing multi-pinhole NRSPECT on a stationary dual detector SPECT system will provide a lowcost upgrade path leading current users of rotational SPECT systems toward acceptance of dedicated multi-pinhole NRSPECT methodology for imaging the heart.