High-speed scanning of planar images showing ¹²³I-MIBG uptake using a whole-body CZT camera: a phantom and clinical study

A planar MIBG phantom (custom-made device, Taisei Medical Co., Ltd., Osaka, Japan; Hokuriku Yuuki Co., Ltd., Kanazawa, Japan) for normal and disease models, which was composed of acrylic plates and parts representing organs such as the heart, mediastinum, lung, and liver, was used (Additional file 1: Figure S1A). As all organ parts were connected as one compartment, no adjustment of radionuclide concentration for each organ part was required. This phantom has been used in several institutions, and we prepared it in the same manner as in previous studies [15, 16]. The thickness of the vacant space at heart/mediastinum was 30/10 and 25/15 mm in the normal and disease models, respectively [16] (Additional file 1: Figure S1 B and C).

Data from 36 consecutive patients with suspected or known PD or DLB who underwent MIBG myocardial innervation imaging using the Discovery NM/CT 670 CZT between July 25 and November 31, 2017, were retrospectively reviewed.

Using the Discovery NM/CT 670 CZT camera, anterior and posterior planar imaging was performed after injection of 55 MBq of ¹²³I-MIBG for the phantom model. In the clinical study, anterior planar imaging was performed 15 min (early) and 3 h (delayed) after intravenous injection of 111 MBq of ¹²³I-MIBG. The Lister tool on the Xeleris 4.0 workstation (GE Healthcare) allowed post hoc reframing of already acquired data with various parameters such as energy window widths and collection times, without additional radioactive tracer administration or imaging acquisition. All images were collected with matrices of 512 × 512 and at an energy peak of 159 keV.

In phantom studies, reference images with a collection time of 300 s were acquired, and these images were reframed with different energy window widths (symmetrical, ± 2.5%, 5.0%, 7.5%, 10%, 20%, and 40%) (Fig. 1a). Subsequently, images with a fixed window width (± 5.0%) were reframed with different collection times (25, 50, 100, and 200 s) (Fig. 1b). The regions of interest (ROIs) were set manually on the heart (52.5 cm², oval-shaped) and mediastinum (10.5 cm², square-shaped) in the anterior and posterior planar views, and the HMR values were calculated using the averaged radioactive counts in these ROIs (Fig. 1c) [17]. A schema of the settings for energy peak and window width and radioactive counts is shown in Fig. 1d. To compare properties between the CZT-based and Anger cameras, images with a fixed window width (± 7.5%) were obtained with different collection times (25, 50, 100, 200, and 300 s) using Picker Prism 2000 equipped with low-energy ultra-high-resolution (LEUHR) and medium-energy general-all-purpose (MEGAP) collimators.

In clinical studies, using a fixed energy peak and window width (159 keV ± 5%), images with the reference collection time of 300 s were first acquired and images with different collection times (25, 50, 100, and 200 s) were subsequently reframed. The HMR was calculated in the same way as in the phantom studies but using only anterior views.

Continuous and discrete variables are reported as mean ± standard deviation (SD). Categorical variables are reported as percentage (%) values. Normality was evaluated using the Shapiro-Wilk test. Continuous values were compared between groups using analysis of covariance. In phantom studies, coefficient of variation (CV) values was calculated using the average and SD counts for ROIs at the heart and mediastinum to evaluate the appropriate energy window width (Fig. 1c). In clinical studies, using intra-class correlation (ICC; two-way mixed-effects model, for consistency) coefficient statistics and Bland-Altman analysis with limit of agreement, the correlation and bias between HMRs in images with the reference collection time of 300 s and those with other collection times (25, 50, 100, and 200 s) were evaluated. Statistical analyses were performed using JMP Pro 11.2.0 (SAS Institute Inc., Cary, NC, USA) or SPSS statistics software (version 25, IBM Corp., Armonk, NY, USA), as appropriate.

Myocardial sympathetic imaging using the conventional Anger camera has been performed in patients with heart failure [1] or LBD [4, 5]. Recent studies have demonstrated the use of the cardiac-dedicated CZT-based cameras D-SPECT and Discovery 530 C for myocardial innervation imaging [10, 11]. In our phantom study, there was an apparent contrast between HMRs in the disease models and the normal models when the energy window was set narrow at ± 2.5%, although this setting implied heterogeneity of the radioactive counts within ROIs, as represented by the high CV values (Fig. 2). This ‘trade-off’ phenomenon was relieved with energy window widths of ± 5%, where the disease-to-normal contrast was still prominent (0.49 in the anterior views) while the CV reduced (0.13 for the disease and 0.12 for the normal model in the anterior views). These preliminary findings led us to set the energy window width at ± 5% in the clinical study, although previous studies on cardiac-dedicated CZT scanners suggested beneficial effects of asymmetrical window settings [10, 11].

In fact, our clinical study demonstrated significantly lower HMRs in patients with clinically diagnosed LBD in comparison with those in patients with other neurologic diseases, suggesting the feasibility of this new modality for differentiating LBD using HMRs.

The currently commercially available cardiac-dedicated CZT-based cameras, D-SPECT and Discovery NM 530 C, have been known to shorten the collection time by a factor of 3–5 in myocardial perfusion imaging using ^99mTc- or ²⁰¹Tl-tracers in comparison with Anger cameras [13, 14, 18]. However, similar evidence of time reduction in myocardial innervation imaging is lacking. Especially, findings showing time reduction using the planar images of the Discovery NM/CT 670 CZT have not been published, except for those involving bone scanning [19]. The reported acquisition time for myocardial innervation imaging with ¹²³I-MIBG using these cardiac-dedicated CZT detectors was approximately 10 min [20, 21], similar to that required by Anger cameras [11, 21, 22]. Our study demonstrates that judging from the ICC analysis, HMRs obtained from images with a collection time of 50 s and longer showed good consistency with those obtained with the default 300 s. Meanwhile, as the collection times became shorter, the difference from 300 s became greater along with wider range of 95% limits of agreement in the Bland-Altman analysis. Additionally, we performed further analysis (Additional file 1: Figure S2) that investigated correlation between the HMRs (reference) and the difference of HMRs (300 s − other collection time). As a result, in the early phase collected with 50 s, there was a significant correlation (r = − 0.35, p = 0.035) between the reference HMRs and the difference of HMRs (300 − 50 s). Thus, disease-to-normal contrast of HMRs might have been preserved with the collection time of 50 s; however, because there were considerable discrepancies of HMRs between 300 and 50 s, such a time reduction cannot be clinically acceptable at this stage. Further studies including regression analysis in larger cohort are needed to elucidate clinical feasibility of such a short collection time. This time reduction for myocardial innervation imaging may be beneficial, because completing the imaging examination may be occasionally challenging for patients as they may suffer heart failure, tremor, or dementia. Additionally, the potential time reduction may also allow administration of lower isotope doses [6‐9, 23].

In our clinical study, the HMRs of the patients without diagnosed LBD were lower (early phase 2.21, delayed phase 2.08) than those in a previous study using an Anger camera; the HMRs in the early and delayed phases obtained from subjects with low likelihood of cardiac disease (i.e. normal subjects) were 2.39 and 2.49 for low-energy collimators and 2.76 and 3.01 for medium- or low-medium-energy collimators, respectively [24]. Discrepancy between the values obtained from Anger cameras and D-SPECT was also considered; thus, a correction method for standardisation of HMRs has already been reported based on phantom experiments [21]. Thus, HMR may vary depending on the type of collimators. In the Discovery NM/CT 670 CZT, the wide-energy high-resolution (WEHR) collimator was used, in which septal penetration generated from high-energy photons (529 keV) may scatter on mediastinal area, resulting in reduction in the HMR, as observed in the low-energy collimators. Our preliminary phantom study (Fig. 4) supports this speculation; the HMRs obtained from CZT-based camera were lower than those from the Anger camera with MEGAP collimators in the normal model. These results indicate the requirement for future prospective, multi-centre studies for standardisation of the HMRs measured by CZT-based detectors equipped with WEHR for whole-body examination.

This was a single-centre, observational study with a relatively small number of patients. However, we suggest that the evidence provided by our data was relevant enough to warrant further larger studies on the time reduction in myocardial innervation imaging and the standardisation of HMRs when using the novel CZT detectors for whole-body imaging.