Evolve or perish for coronary calcium imaging

Advanced coronary atherosclerosis is characterized by calcium deposition. Although coronary calcium had been visualized by fluoroscopy and conventional computed tomography (CT), it was not until the introduction of electron beam computed tomography (EBCT) that high-resolution, motion-free images of the coronary arteries could be acquired. In their seminal paper from 1990, Agatston et al.1 describe a methodology to image and quantify calcium in the coronary arteries. Cohort studies demonstrated how coronary calcium varies between genders and races, but progressively increases with age. Coronary calcium correlates with the overall atherosclerotic burden, predicts adverse events with an incremental value over conventional risk factors, and the test can be used for risk restratification of intermediate-risk patients.2 Severe coronary disease is rare in the absence of calcium on CT,3 and although not all plaque is calcified, calcium scanning appears sufficiently reliable to exclude coronary disease in lower-risk patients. Despite the emergence of coronary CT angiography (CCTA), calcium imaging is still performed in conjunction with CCTA at many centres, for risk stratification, gatekeeper to CCTA, or CCTA exam optimization, justified by the only fractional increase in total radiation dose. However, CCTA further evolved. Over the past decade axial and high-pitch protocols, iterative reconstruction and lower roentgen tube potentials have dramatically reduced the radiation dose of CCTA. Currently, routine CCTA can be performed at 2–4 mSv, while contemporary technology allows acquisitions of <1 mSv. In this new reality, the radiation dose of a calcium scan, ∼1–1.5 mSv, is no longer negligible. How is it possible that a relatively low-resolution, non-enhanced CT scan requires more radiation than a high-resolution CT coronary angiogram? The answer lies in the past. The Agatston method was developed within the technical limits of EBCT technology.1 Without rotating mechanical parts, EBCT permitted motion-free imaging of the coronary arteries. The first scanners allowed acquisition of only a single 3 mm slice per cardiac cycle, and 20 consecutive slices in a single breath-hold examination. Empirically, a threshold of 130 HU differentiated best between real coronary calcium and image noise. Also to avoid misinterpretation of noise, Agatston's method required at least two connecting image elements to reach the 130 HU threshold to be considered calcium. Per slice the area of each calcium spot was measured, and multiplied by a weighing factor (1–4) determined by the maximum lesion attenuation. Finally, the scores of all lesions and in every slice were added up as the total calcium score. As the longitudinal scan coverage increased and the detector element sizes decreased with multidetector systems, more quantitative measures were possible, including the calcium volume, the calcium mass (considering the spatial attenuation distribution), and the calcium density. Nevertheless, because of the familiarity and wealth of data acquired using the Agatston score, this methodology remained in general use. Hecht et al.4 investigated a low-dose calcium imaging protocol, through lowering the tube current (mA) by half, and using iterative image reconstruction to compensate for increased image noise. The low-dose scan was performed in addition to a standard calcium scan protocol, which allowed for per patient comparison of scan parameters, radiation doses, image quality, and calcium scores. While the low-dose protocol achieved a 50% dose reduction, image quality did not suffer, in fact signal-to-noise ratios improved. The calcium scores were very similar, but with a small mean underestimation using the low-dose protocol. While this difference did not affect calcium score classification for most individuals, there were two patients in whom a small amount of calcium was no longer detectable using the low-dose protocol. Similar to the study by Hecht et al., other groups have investigated low-mA calcium imaging,5 although dose reduction was less without iterative reconstruction techniques (Table 1). Alternatively, the tube voltage (kVp) can be lowered. Calcium imaging at 100 kVp is well possible, but increases both contrast and noise in the images, necessitating a higher attenuation threshold (147 HU) to derive a comparable Agatston score.9,10 Whether the roentgen tube current or the voltage is lowered, both increase image noise. As demonstrated in the present study, iterative reconstruction techniques reduce image noise. Some researcher, however, reported changes in measured calcium scores depending on the type and degree of iterative reconstruction.6–8 Additionally, excessive iterative reconstruction may smoothen out small calcifications, as demonstrated in this paper.4,7 While this does not affect the overall score substantially, it may become relevant when the calcium score is used in a diagnostic fashion to rule out coronary artery disease. Whether we strive for overall reproducibility or preserved sensitivity largely depends on whether the test is used for risk stratification or to rule out of coronary disease. Summary of contemporary calcium scan dose reduction studies IR, iterative reconstruction. aFive patients excluded because of excessive image noise. Summary of contemporary calcium scan dose reduction studies IR, iterative reconstruction. aFive patients excluded because of excessive image noise. For calcium imaging to survive, the test will need to adapt to a new environment, rather than persisting on historical consistency. Frankly, current calcium imaging is not that reproducible either, and scores vary between examinations, scan protocols, scanner systems, and quantification software.11,12 Likely, calcium scores have drifted over time as scanner performance advanced. Perhaps, it is time to let go of the high tube voltage, the thick cuts, filtered back-projection, the 130 HU threshold, and semi-quantitative scoring. There is an opportunity to reinvent calcium imaging, at a lower dose, using new and perhaps individualized attenuation thresholds, and with more quantitative measures. A change in values, units, or references will cause confusion at first. But like high-sensitive troponin assays, you get used to it in the end. Unless we manage to reduce the radiation dose of calcium imaging, it will become a thing of the past.