Value of automatic patient motion detection and correction in myocardial perfusion imaging using a CZT-based SPECT camera

Abstract

Background: Correction of motion has become feasible on cadmium-zinc-telluride (CZT)-based SPECT cameras during myocardial perfusion imaging (MPI). Our aim was to quantify the motion and to determine the value of automatic correction using commercially available software.

Methods and results: We retrospectively included 83 consecutive patients who underwent stress-rest MPI CZT-SPECT and invasive fractional flow reserve (FFR) measurement. Eight-minute stress acquisitions were reformatted into 1.0- and 20-second bins to detect respiratory motion (RM) and patient motion (PM), respectively. RM and PM were quantified and scans were automatically corrected. Total perfusion deficit (TPD) and SPECT interpretation-normal, equivocal, or abnormal-were compared between the noncorrected and corrected scans. Scans with a changed SPECT interpretation were compared with FFR, the reference standard. Average RM was 2.5 ± 0.4 mm and maximal PM was 4.5 ± 1.3 mm. RM correction influenced the diagnostic outcomes in two patients based on TPD changes ≥7% and in nine patients based on changed visual interpretation. In only four of these patients, the changed SPECT interpretation corresponded with FFR measurements. Correction for PM did not influence the diagnostic outcomes.

Conclusion: Respiratory motion and patient motion were small. Motion correction did not appear to improve the diagnostic outcome and, hence, the added value seems limited in MPI using CZT-based SPECT cameras.

Keywords: Myocardial perfusion imaging (MPI); cadmium-zinc-telluride (CZT); motion correction; patient motion; respiratory motion.

Figure 1

The mean (A) respiratory motion in the cranial-caudal direction and (B) patient motion in all three directions and the overall patient motion as function of scan time. The respiratory motion decreased significantly during the 8-minute acquisition (P = .01), whereas this was nearly significant for the overall patient motion (P = .06)

Figure 2

Relation between both respiratory (top row) and patient motion (bottom row) and the differences between the noncorrected and motion-corrected scans in (A, D) the visual SPECT interpretation, (B, E) the total perfusion deficit, and (C, F) the number of changed segments ≥5%. Neither patient motion nor respiratory motion correlated with the differences in one of the three endpoints (P > .26). Note that no changes in SPECT interpretation were found after correction for patient motion (D). The dashed lines represent the thresholds for which the diagnostic outcome was considered to be influenced

Figure 3

Bland Altman plots of the noncorrected and motion-corrected images for (A, B) the respiratory motion (RM) and (C, D) the patient motion (PM). The left images (A, C) show the differences in total perfusion deficit (TPD) and the right images (B, D) show the differences in segmental uptake values of all 17 segments of all patients. The shaded areas represent the 95% confidence interval limits and the long dashed lines represent the thresholds for which the diagnostic outcome was considered to be influenced