Diagnostic accuracy for CZT gamma camera compared to conventional gamma camera technique with myocardial perfusion single-photon emission computed tomography: Assessment of myocardial infarction and function

Study Population and Design

The study protocol was approved by the Regional Ethics Committee at Lund University (LU2013/550 and LU2013/4010). Patients were included in two ways. (1) Patients clinically referred for CMR imaging, where CMR images showed evidence of ischemic scar were asked to participate in the study. The patients were examined with MPS at rest and images were acquired in two gamma cameras, a CZT gamma camera and a conventional gamma camera. Out of 47 included patients, 7 were excluded because SPECT data for both gamma cameras could not be obtained due to intermittent technical problems with the scanner table on the conventional gamma camera, one was excluded because of inadequate LGE-CMR image quality, one was excluded because CMR was performed during the acute phase of the MI and two were excluded because of presence of left bundle branch block which is known to possibly affect the MPS image uptake pattern.11 Thus, 36 patients could be used for image analysis of both MPS and CMR. (2) In addition, a subset of the patients was recruited from another study (the MYOMER study), in which patients clinically referred for an elective coronary angiography (CA) because of known or suspected chronic coronary syndrome were included. The goals with the MYOMER study were to study myocardial perfusion imaging before and after CA, with or without percutaneous coronary intervention. From the MYOMER study, 37 patients were examined with CMR and MPS at rest with image acquisition in both gamma cameras before the CA examination, and therefore could be included in the current study. Thus, in total the study population consisted of 73 patients with two MPS image acquisitions each resulting in 146 datasets that were evaluated. Patient charts were reviewed for patient characteristics and to exclude any cardiac adverse event, coronary revascularization or changes in cardiac medication occurring between the CMR and MPS examinations.

MPSImage acquisition

Patients were examined at rest and injected with a weight adjusted activity of 4 MBq·kg−1 of 99mTc-tetrofosmin (GE Healthcare) (356 ± 69 MBq). Image acquisition was performed 45-60 minutes after the injection. Each patient was examined in both supine and prone position and in both a cardiac dedicated CZT gamma camera (Discovery NM 530c, GE Healthcare) and a cardiac dedicated conventional gamma camera (Ventri, GE Healthcare). There was no systematic order in which gamma camera was used for the first and second image acquisition, since image acquisition of the study patients had to be accommodated to the clinical flow of patients.

The acquisition time on the CZT gamma camera was 480 seconds. The images were reconstructed with a Maximum Likelihood Estimation Method (MLEM) algorithm, 40 iterations; Green OSL regularization α parameter of 0.51 and a β of 0.3 and post filtered with a Butterworth filter with a cut-off frequency of 0.37 and a power of 7. For the conventional gamma camera the examination was performed with the detectors in L-mode. Sixty projections were acquired in a total angular range of 180° with a stop condition of 25 seconds per projection. The conventional gamma camera images were reconstructed with a resolution recovery OSEM algorithm (Evolution, GE Healthcare) using 12 iterations and 10 subsets and post filtered with a Butterworth filter with a cut-off frequency of 0.4 and a power of 10. All reconstruction parameters used followed recommendations from the manufacturer. The reconstructed images were reformatted to the standard cardiac axis format (short-axis, vertical long-axis and horizontal long-axis). ECG-gated image acquisition using 8 frames per cardiac cycle was performed for all supine acquisitions. ECG-triggering failed in two acquisitions for the conventional gamma camera, due to poor ECG signal. Attenuation correction was not applied.

MPS image analysis

All MPS images were analyzed using the software Segment, version 2.2 (Medviso AB, Lund, Sweden) and QGS/QPS, version 2015.6 (Cedars-Sinai, Los Angeles, USA). For LV volumes, EF and mass the software were used following recommendations from the manufacturers. Briefly, both reconstructed static and gated images were analyzed by fully automated LV segmentation algorithms. LV end-diastolic volume (LVEDV), LV end-systolic volume (LVESV), LV stroke volume (LVSV) and LVEF were calculated from the gated images and LV mass (LVM) was calculated from the static images. Manual correction was performed if the automatic segmentation was obviously wrong. For assessment of MI, gated and summed MPS images, acquired in supine position, were loaded into the QGS/QPS software. One experienced observer, blinded to patient data, visually evaluated the images in random order as previously described.3 Briefly, a perfusion defect in the summed images with decreased wall thickening in the gated images was reported as MI. If the observer felt uncertain after evaluating the images acquired in supine position, the summed images acquired in both supine and prone position were used. Infarcts were located to the left anterior descending artery (LAD) territory (anterior, septal and/or apical parts of the LV) or to the left circumflex artery/right coronary artery (LCx/RCA) territory (lateral and/or inferior parts of the LV). Additionally, regional myocardial tracer uptake in each LV segment according to the standardized 17 segments model12 was quantified using a 5-point scale ranging from 0 (normal uptake) to 4 (absent uptake), where regional motion according to the gated images was taken into account. Thus, for a segment to be scored as reduced uptake, regional motion in that segment would have to be affected. If uptake was judged to be reduced but motion in that segment was judged to be normal, the score was set to 0. The total score of the left ventricle at rest, summed rest score (SRS) was calculated. Twenty cases were evaluated twice and by a second observer to calculate intra- and interobserver variability for infarct detection. For both the CZT and the conventional gamma camera, epi- and endocardial borders were derived from automated delineation provided by the MPS software. Therefore, intra- and interobserver variability for functional parameters by MPS were not assessed.

CMRImage acquisition

CMR imaging was performed on a Philips Intera CV (Best, The Netherlands) for seven patients, on a Siemens Magnetom Aera (Erlangen, Germany) for 63 patients and on a Siemens Magnetom Avanto (Erlangen, Germany) for three patients. All subjects were placed in supine position. Cine short-axis gradient-recalled echo images covering the left ventricle were acquired using a balanced turbo field echo (bTFE) sequence: slice thickness = 8 mm, field-of-view = 340 mm, TR = 3.14 ms, TE = 1.58 ms. Three cine long-axis images (2-, 3- and 4-chamber views) were acquired using the same sequence. Approximately 15 min after intravenous administration of an extracellular gadolinium-based contrast agent (gadoteric acid, Gd-DOTA, 0.2 mmol·kg−1, Guerbet, Gothia Medical AB, Billdal, Sweden) an inversion-recovery (IR) sequence was used to acquire late gadolinium enhanced (LGE) images in the corresponding planes as for the cine images. Typical LGE sequence parameters were: slice thickness = 8 mm, TR = 3.9 ms, TE = 1.2 ms, in-plane resolution = 1.5 × 1.5 mm and flip angle = 15º with acquisition every heartbeat. The inversion time, typically 250-350 ms, was manually adjusted to null the signal from remote myocardium.

CMR image analysis

All CMR images were analyzed using the software Segment, version 2.2. The endo- and epicardium of the LV were manually delineated in the cine short-axis images in both end-diastole and end-systole by two observers in consensus. The LV end-diastole and end-systole were defined as the time frame with the largest and the smallest LV blood pool volume, respectively. Trabecular and papillary muscles not contiguous with the myocardial wall were excluded, thus included in the LV cavity volume. The endo- and epicardial borders were adjusted between end-diastole and end-systole to accomplish the same LVM in both time frames. Based on the LV delineation, LVEDV, LVESV, LVSV and LVEF were calculated. LVM was calculated as the muscle volume between the endo- and epicardial delineations, multiplied by the density of the myocardium (1.05·g·mL−1).13 Assessment of MI was performed on the LGE-CMR images, where hyperenhanced regions extending from the LV endocardium in two perpendicular imaging planes according to typical coronary artery territories, were considered MI. MI’s were visually located to the LAD or the LCx/RCA territory. MI size was quantified using a semi-automatic method, the EWA algorithm, as previously described with manual corrections if needed.14 Infarct transmurality was quantified as the infarct extension measured from the endocardium to the epicardium, both per segment and for the over-all infarct (mean transmurality).

Statistical Analysis

Data are presented as mean ± SD or median (interquartile range 25%-75%) unless otherwise stated. All statistical calculations were performed using either Prism 7.04 (GraphPad Software, San Diego, CA, USA) or Microsoft Excel 2013 (Microsoft Corporation, Redmond, WA, USA). LV volumes, EF and mass by MPS and CMR were compared using Student’s t-test. The absolute differences between MPS and CMR for LV volumes, EF and mass were investigated with modified Bland-Altman analysis, using the reference method CMR on the x-axis and the absolute difference between MPS and CMR on the y-axis. A P-value of < .05 was considered to indicate statistical significance.

留言 (0)

沒有登入
gif