TO THE EDITOR:
We read with interest the article by Liu et al. in the August 2002 issue of the Journal (1). To reiterate, SPECT images acquired with a 180° orbit may have significant erroneous inhomogeneity and overestimate defect size if the target object is off the center of the orbit. We present 2 cases of patients who recently underwent 180° acquisitions that led to false-positive findings for the anteroseptal wall of the left ventricle.
Our routine procedure for dual-isotope (rest 201Tl and stress 99mTc-tetrofosmin) cardiac gated SPECT uses a 3-head gamma camera (Prism; Picker) to acquire 360° SPECT data; however, following the guidelines of the American Society of Nuclear Cardiology (2), we use a circular orbit and an anterior 180° of SPECT acquired data to process SPECT images. We frequently find a small area of decreased uptake in the anteroseptal or inferolateral wall in the short axis of the SPECT image. This finding occurs on both 201Tl and 99mTc-tetrofosmin SPECT images. Here are 2 recent examples.
The first case was a 41-y-old man with a history of alcohol abuse 4–5 y ago who complained of atypical chest pain with and without exertion over the past 2 mo. He was referred for dipyridamole 99mTc-tetrofosmin and rest 201Tl-chloride cardiac gated SPECT. Moderate enlargement of the left ventricle and slightly decreased activity in the anteroseptal area were seen in the short axes of both the 99mTc-tetrofosmin and the 201Tl cardiac SPECT images. These SPECT acquisitions depended on 180° data. In SPECT images obtained using 360° data, the area of decreased uptake in the anteroseptal wall was no longer seen in either the 99mTc-tetrofosmin or the 201Tl short axes. Left ventricular ejection fraction was 51% (normal ejection fraction is ≥50%).
The second case was a 59-y-old man with prostate cancer who was scheduled for radical perineal prostatectomy and was referred for dipyridamole 99mTc-tetrofosmin and rest 201Tl-chloride cardiac gated SPECT for presurgical evaluation. The patient had a history of hypertension and coronary artery disease and had undergone angioplasty in the past. Imaging was performed using a 3-head gamma camera. Slightly decreased uptake in the anteroseptal wall was more apparent in the short-axis 201Tl SPECT images than in the 99mTc-tetrofosmin SPECT images. On SPECT images obtained using 360° data, the hypoperfusion in the anteroseptal wall had almost disappeared. Gated SPECT showed normal wall motion of both ventricles. Left ventricular ejection fraction was 61%.
Our 2 cases concur with reports that SPECT images acquired with a 360° orbit may provide more accurate quantitative information. Inhomogeneity may occur with 180° acquired SPECT data. The use of a circular orbit and an anterior 180° (right anterior oblique to left posterior oblique) acquisition orbit has been standardized by the American Society of Nuclear Cardiology (2); however, because 360° data are available from our 3-head camera, our policy is to use such data whenever an area of hypoperfusion in the anteroseptal wall is suspected. SPECT images acquired with a 360° orbit may provide more accurate quantitative information and are less likely to be affected by artifacts.
REFERENCES
REPLY:
We appreciate the opportunity to address some of the issues raised by Dennis Kirch and colleagues regarding our original article (1). The major concern mentioned in the Kirch letter was the “wavy” circumferential count profiles of the “normal” phantom in the center position using 180° and 360° acquisition orbits. The focus of our study was to investigate in a systematic manner the effects of 180° image acquisition and phantom position on image uniformity. Although we were also intrigued by the presence of the wavy artifact, we believe that it did not affect the validity of our findings and conclusions, since all images were acquired using the same SPECT camera. Furthermore, as mentioned in our article, we also performed complementary experiments using a different triple-head SPECT camera. The findings using this different camera were entirely consistent with those presented in our article (1).
We certainly agree with Kirch’s comments that SPECT imaging is a complex process, in which mechanic deficiencies may contribute to image distortion.
Regarding systematic error, although the tasks specifically suggested by Kirch were not executed in our study, the SPECT camera has been maintained and serviced scrupulously according to the vendor’s specification. Center-of-rotation offset is checked quarterly. Quality control for detector uniformity and image resolution is performed by technologists daily.
Regarding experimental error, for quantification of the SPECT studies, the basal portions of the phantom close to the wood frame were excluded to minimize potential scatter and attenuation from the wood. The wavy pattern was present in all apical, midventricular, and basal slices. The fact that the pattern was slightly greater when the phantom was angled was in accordance with our assumption. Thus, we believe that the wavy artifact was most likely caused by imperfect positioning of the phantom and deviation from the centerline. We agree that replacing the wood frame with a paper or fabric cradle and positioning the phantom with a laser device might be helpful to eliminate the potential effect of scattering material. Although, theoretically, using an asymmetric 99mTc energy window may improve the primary-to-scatter ratio as suggested by Kirch, Devous et al. showed that in practice this step did not significantly improve image quality (2).
Regarding analytic error, the circumferential count profile generated by the WL-CQ software (Eclipse Systems), not being based simply on a single maximal pixel value, was contrary to what was assumed by Kirch. Instead, a bilinear interpolation scheme of averaging 4 nearest radial and circumferential neighbors of the maximal pixel was incorporated into our data sampling. Thus, a smoothing strategy similar to that Kirch suggested had been adapted into our count profile generation, although it was not described in detail in our previous publications (3,4).
We also appreciate the comments of Wei-Jen Shih and colleague regarding differences in clinical image uniformity associated with 180° and 360° acquisitions. Clinical SPECT images acquired with a 360° orbit may provide more accurate quantitative information, whereas nonhomogeneity may occur with 180° acquired SPECT data. These clinical observations support our findings of the phantom study. We have also observed the same phenomenon reported in our patient population. In fact, these observations prompted our conducting of this phantom study. It is important to realize that, because of differences in patient body habitus, artifacts due to 180° acquisition may be more or less prominent in individual patients.
Ultimately, we would like to emphasize that the purpose of this phantom study was to answer a fundamental question: Does 360° full-image acquisition improve quantitative SPECT accuracy in terms of count profile homogeneity and myocardial perfusion defect size? Our results suggest that central positioning of the target imaged and 360° acquisition are important parameters for optimal image quality. Although the phantom we used appears to be “simplistic,” it allowed us to investigate a single variable in isolation. Increased nonhomogeneity and inconsistent defect size quantification appear to be caused primarily by the depth-dependent point-spread function of SPECT imaging systems. Nonhomogeneity and inconsistent quantification are suboptimal when a 180° imaging orbit is used.
As Kirch mentioned, special effort in positioning patients for clinical SPECT imaging is needed to ensure diagnostic comparability and reproducibility. On the basis of our clinical observation and phantom study, we have modified our practice to perform only 360° image acquisitions. We find that this approach improves image uniformity.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.