Additional Diagnostic Value of Integrated Analysis of Cardiac CTA and SPECT MPI Using the SMARTVis System in Patients with Suspected Coronary Artery Disease

Study Population

Seventy-one patients who underwent cardiac CTA and SPECT MPI at the Leiden University Medical Center (Leiden, The Netherlands) were randomly selected. After application of exclusion criteria (Fig. 1), 17 patients were included in our study; images from an ICA procedure performed within a 90-d period were available for 7 of them. Patient characteristics are presented in Table 1. The institutional review board approved this retrospective study, and the requirement to obtain informed consent was waived.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Patient inclusion and exclusion criteria. CABG = coronary artery bypass grafting; PCI = percutaneous coronary intervention.

SPECT MPI

SPECT MPI was performed using a 2-d protocol starting with the stress acquisition. The patients underwent bicycle ergometry or, when contraindications were present, adenosine or dobutamine infusion to induce stress. The radioisotope (500 MBq of ^99mTc-tetrofosmin) was injected at peak exercise, at 3 min after starting adenosine perfusion, or at peak heart rate during dobutamine infusion. For both stress and rest scans, the images were acquired 1 h after radioisotope injection. A triple-head camera system (CGA 9300; Toshiba) and a low-energy high-resolution collimator were used. Electrocardiographic gating was performed at 16 frames per cardiac cycle, with a tolerance window of 50%. No attenuation or scatter correction was applied.

An experienced nuclear physicist, masked to both CTA and ICA results, analyzed the scans using the Corridor4DM software package (version 6.1; INVIA Solutions) (18). SPECT MPI images were interpreted using oblique slices, polar maps, and quantitative or functional values. The SPECT MPI interpretation was summarized into a report, using the following guidelines (19): the observer graded each of the 19 myocardial segments as being normal (no perfusion defect) or abnormal (reversible or fixed defects) and indicated the extent of myocardial infarction or ischemia.

CTA

Five patients were scanned using a 64-slice CT scanner (Aquilion 64; Toshiba Medical Systems Corp.), and the remaining 12 patients were scanned using a 320-slice CT scanner (Aquilion ONE; Toshiba Medical Systems Corp.). When the heart rate was higher than 65 beats/min, additional oral β-blockers (metoprolol, 50 mg, single dose, 1 h before the scan) were provided when tolerated. A prospectively triggered coronary calcium scan (unenhanced CT scan) was performed before the CTA acquisition. CTA images were acquired with a collimation of 64 × 0.5 mm for the 64-slice scanner and 320 × 0.5 mm for the 320-slice scanner, a tube rotation time of 400 ms, and a tube current of 300 mA at 120 kV for patients with a body mass index of less than 30 kg/m². If a patient had a higher body mass index, tube current was increased to 350 or 400 mA at 135 kV. The acquisition of imaging was prospectively triggered at 75% of the R–R interval. Between 80 and 110 mL of nonionic contrast material (Iomeron 400H; Bracco Atlanta Pharma) were administered with a flow rate of 5 mL/s depending on the total scanning time. The timing of the scan was determined using automated detection of peak enhancement in the aortic root. Th acquisition was conducted during an inspiratory breath-hold of approximately 10 s.

A single experienced reader, masked to both SPECT MPI and ICA results, analyzed the scans using the syngo.via workstation (Siemens Healthcare). CTA images were interpreted using transaxial image stacks and curved multiplanar reformatted images. The CTA interpretation per coronary segment (American Heart Association Model (20)) was summarized into a report, using the following guidelines (21): the observer reported the stenosis location (origin, proximal, mid, distal, or end), the stenosis severity (mild, moderate, severe, or occluded), and the stenosis plaque type (noncalcified, mixed, or calcified).

Quantitative Coronary Angiography (QCA)

ICA was performed according to standard medical practice. One experienced cardiologist unaware of the CTA and SPECT MPI scoring results performed QCA on the 7 available angiograms. All coronary segments were identified and analyzed using the modified 17-segment American Heart Association classification. Segments were visually classified as normal (smooth parallel or tapering borders, visually ≤ 20% narrowing) or as having coronary obstruction (visually ≥ 20% narrowing); the stenoses in these last segments were quantified by a validated QCA algorithm (CAAS; Pie Medical) (22). Stenoses were evaluated in the worst angiographic view and classified as significant if the lumen diameter reduction exceeded 50%.

SMARTVis: A Software-Based CTA/SPECT MPI Fusion System

In this work, we extended the SMARTVis software-based system that we introduced in 2012 (23) to fuse CTA with SPECT MPI data. An overview of the CTA and SPECT MPI processing and fusion is given in Figure 2. The SMARTVis system provides comprehensive 2-dimensional (2D) and 3D fused visualizations of the anatomic and functional information for relating coronary stenoses and perfusion defect regions. The coronary artery tree extracted from CTA can be projected onto the 2D stress–rest polar map, and similarly, the perfusion information visualized on a 3D stress–rest polar map can be fused with a 3D model of the heart and its coronary artery tree. Furthermore, the SMARTVis system provides a list of automatically detected and quantified coronary artery stenoses (24). To further assist the user in assigning a culprit lesion to a specific perfusion defect, a distance-based estimation of the patient-specific coronary perfusion territories is provided. Last, the 2D and 3D polar map views are synchronized with the CTA stenosis findings and images.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Overview of image processing performed on CTA and fusion of CTA/SPECT MPI. Dashed box corresponds to semiautomatic process, whereas solid boxes correspond to fully automatic processes. Previously published methods (24–26) were used to detect and quantify coronary artery stenoses on CTA, quantify coronary calcium, and obtain cardiac chamber shapes from CTA. SPECT MPI left ventricle shape was automatically provided by Corridor4DM software, as were landmark points indicating septal and apical positions. LV shapes and landmark points were subsequently used to align CTA and SPECT MPI data by applying iterative closest-point algorithm. 3D fusion was qualitatively validated by visual inspection of correctness of alignment. Endo/epi = endocardium/epicardium.

Study Design

The additional diagnostic value of the SMARTVis system to assess CAD was investigated through a case-study evaluation involving 4 experts from 2 medical centers (Erasmus Medical Center, Rotterdam, The Netherlands, and Leiden University Medical Center, Leiden, The Netherlands). An overview of the study design is presented in Figure 3.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Overview of multicenter evaluation study design. First, structured reports were created for CTA and SPECT MPI of 17 patients. QCA analysis was performed for 7 patients. Treatment strategy (i.e., medical treatment or revascularization of specific coronary segments) was further derived from QCA and SPECT MPI findings and served as reference standard. Because guidelines recommend proof of ischemia before revascularization of coronary stenoses (11), expert considered medical therapy indicated for significant lesions detected on QCA that resulted in infarction and no further complaints. During individual sessions, 4 experts examined 17 patients and performed, first, side-by-side analysis using CTA and SPECT MPI reports and, second, integrated analysis using SMARTVis fusion system in addition to reports. Side-by-side analysis was performed 2–5 wk before integrated analysis to minimize chance of recalling patient cases; patients were analyzed in different order, and expert analyzed several other cases in elapsed time.

During the analyses, for each of the 17 coronary artery segments that had at least one stenosis of at least 20% on CTA, the observer first had to indicate the correlation between stenoses and perfusion defects. The stenoses in the considered coronary segment could either be related to a myocardial region showing a perfusion defect on SPECT MPI or be considered hemodynamically not significant (i.e., not inducing a perfusion defect). Subsequently, the observer had to indicate the most appropriate therapeutic decision: medical therapy or revascularization of specific coronary segments.

Analysis

The revascularization decision strategy and target vessel selection of each observer were compared both with the other observers’ decisions and with the reference decision derived from interpretation of QCA and SPECT MPI. The diagnostic performance of CTA, SPECT MPI, and their fusion was compared on a per-vessel basis to determine the therapeutic-decision agreement, as well as the sensitivity and specificity.

For all 17 patients, an interobserver therapeutic-decision agreement percentage was computed per patient p as follows:

with v the main arteries (right coronary artery [RCA], left anterior descending coronary artery [LAD], left circumflex coronary artery [LCX], intermediate branch) and ω_v the observer therapeutic agreement for vessel v.

For the subset of 7 patients who underwent ICA, a therapeutic-decision agreement percentage with respect to the QCA/SPECT MPI decision was computed per patient p as follows:

with the QCA/SPECT MPI decision. Also, the sensitivity and specificity for revascularization of a coronary artery were computed.