The D-SPECT demonstrated significantly improved count rates with decreasing activity volume, a feature not previously encountered in nuclear medicine cameras. This camera feature is explained by the D-SPECT’s internal mechanics. According to Spectrum Dynamics (personal communication), each of the D-SPECT's nine DCs initially acquires images at 120 steps across a 110° arc, with equal time spent at each position. After the pre-scan, the computer adjusts DC movement so that 60% to 70% of angles view the LV as defined by the technologist-rendered ROI and 30% to 40% view the remainder of the chest. We can describe the 60% to 70% left ventricular views as the central portion of the field of view (CPFOV). As the heart size decreases, the DC angular intervals in the CPFOV become finer and more closely stacked causing increased overlap of consecutive DC images, at the expense of slightly coarser peripheral sampling. Thus, the virtual detector contracts or expands the CPFOV in response to heart size. Denser oversampling of smaller hearts increases cardiac count rate. The increase in sampling density and count rate with CPFOV shrinkage resembles ‘thickening’ of the detector as it contracts, like a muscle or rubber band. Another analogy that may capture this effect is a searchlight gliding over a field. A smaller search area enjoys improved illumination because the searchlight dwells longer at each location.
In contrast to the D-SPECT, fixed field of view cameras (D530c and A-SPECT) did not demonstrate similarly improved count rates with decreased column height. In fact, D530c count-rate increased slightly with increased column height. Further study may help assess whether this occurred because more of the activity was perpendicular to the pinholes with increased column height. The A-SPECT's count rates showed a small but statistically significant increase with decreasing column height. The increase is unlikely to be important clinically and is likely due to decreased self-attenuation by the smaller column. Although decreased self-attenuation may also increase D-SPECT count rates, the small scale of this effect cannot account for the much larger increase in count rate observed for the D-SPECT with decreased column height. It should be noted that, because these cameras have different resolution and efficiency characteristics, count rate does not reflect the resulting contrast to noise ratios that can be achieved for a given acquisition time.
Beyond the D-SPECT's adaptable virtual detector, there are supplemental causes for the rapid and high quality scans of small hearts that originally lead to this investigation. Images of smaller hearts are less noisy due to a higher count density, since total LV counts are shared by fewer milliliters or voxels. Additionally, small hearts are associated with smaller patients [12]. Decreased soft tissue attenuation in small patients allows more photons to escape the body and reduces scattered photons that deliver incorrectly assigned points of origin. This leads to higher count rates and better resolution. Voxel count density and attenuation affect all camera systems equally and unrelated to the D-SPECT’s increased count rate in small hearts identified in our phantom study.
CZT cameras possess small fields of view that are quite sensitive to heart position [13]. Smaller patients' hearts are more ideally positioned in the field of view. Additionally, closer proximity to the detector should improve image resolution and count rate for both cameras. Because only CZT cameras scan in direct chest contact, the improved heart-detector proximity due to decreased chest size in smaller patients is proportionately much more significant for CZT than A-SPECT cameras. For the D-SPECT, the center of the circle likely allows all nine DCs to view the heart from the closest average distance. Although D-SPECT uses a parallel hole collimator and Anger camera parallel hole collimator efficiency does not change with distance, D-SPECT sensitivity should decrease with increased distance from the heart. This is because of the size difference between A-SPECT and D-SPECT collimators. The Anger cameras have large detectors where the sensitivity lost per individual collimator-hole is offset by an equal increase in the number of collimator holes viewing an object as it recedes. Since the D-SPECT's 4-cm wide DCs do not span the average internal end systolic LV diameter [14], the number of collimator holes exposed to the heart does not increase in step with the decrease in individual collimator hole sensitivity as the object recedes. For the D530c, the crosshairs location presumably minimizes average myocardial displacement from the 19 pinhole apertures, since pinhole sensitivity falls with increasing angle and distance [15].
From a resource management perspective, extra high quality scans are ‘overexposed’ and present an opportunity to decrease radiopharmaceutical dose and/or scan-time. Although all cameras register higher count rates in smaller patients, only the D-SPECT capitalizes on this by shortening scan duration, thanks to its scan strategy that targets LV counts rather than a predetermined exposure time. The D-SPECT scan times could potentially be shortened even further for smaller hearts by targeting count density rather than total counts. LV volume might be estimated from the pre-scan. BMI input could further refine count-targets because improved heart detector proximity and less scatter in smaller patients could allow scan completion with lower than normal count-density. Currently, the Spectrum Dynamics recommends outlining the pre-scan LV with generous margins. As this study indicates, a tighter LV margin will increase count-rate by concentrating the CPFOV. Further study may reveal if tighter ROIs can be drawn to increase scan speed without degrading image quality.
In locales where heart sizes are smaller due to ethnic variations [16], this may mitigate slower scan speed associated with D-SPECT versions employing fewer DCs.
Other cameras could potentially personalize scan time without adding a time-consuming pre-scan by measuring LV count rate during early acquisition, provided that the exposure time adjustments described above could be completed well before the scan would normally end. D530c is particularly well suited for this because its inherent full-time SPECT configuration permits scan termination at any instant.
Limitations
While different sized cardiac phantoms would be ideal, they are not available. Initially we attempted to use different sized hollow spheres to approximate the size and shape of the left ventricle. However, small random variations in dose-calibrator sphere-activity assays marred count-rate/megabecquerel comparisons between the different sized spheres. Since the syringe set only required repositioning five times for each camera, and care was taken to place it at the same spot each time, variability due to source location was minimal. Location was undefined only for the A-SPECT, but is unlikely to be relevant. The D-SPECT ROI was allowed to vary with object size as it would in practice, but its shape was standardized to avoid differences in ROI between repeat scans.