Photon-counting computed tomography (PCCT) is a “revolutionary new imaging technique,” according to Judith van der Bie, MSc, of the Department of Radiology and Nuclear Medicine at Erasmus Medical Centre in Rotterdam, The Netherlands.
On Sunday, van der Bie presented the results of a study that compared the use of PCCT with other methods in evaluating coronary calcifications.
van der Bie
“With conventional energy-integrating computed tomography, the assessment of coronary calcifications is troublesome, due to blooming artefacts and partial volume effects for small calcifications,” she said. “PCCT allows for improved spatial resolution and the performance of spectral post-processing. One of these features consists of algorithms that can extract the signal of contrast agents from contrast-enhanced scans.”
The team conducted a study that focused on two algorithms: virtual-non-contrast (VNC) reconstructions and pure calcium (PC).
“We assessed the performance of both algorithms to compute Agatston coronary calcification (CAC) scores and compared the scores to true-non-contrast (TNC) enhanced scans,” she said. “Accurately defining CAC scores by these algorithms could obviate the need for a true-non-contrast enhanced acquisition, and could therefore lead to dose reduction for patients.”
Measuring the Accuracy of PCCT
Study participants included 88 patients (mean age, 59 years; 69% male) who had a TNC coronary calcium scan and then underwent computed tomography angiography (CCTA) on PCCT. The researchers virtually removed the signal of the contrast agent and then computed the VNC and PC spectral reconstructions using the CCTA datasets. They then determined coronary Agatston scores on all TNC, VNC and PC images.
The investigators classified patients by CAC using the following categories:
- None (CAC=0)
- Minimal (CAC>0 – 10)
- Mild (CAC>10 – 100)
- Moderate (>100 – 400)
- Severe (CAC>400)
Next, they analyzed the overall agreement of the reconstructions using Bland-Altman plots and the level of matching classifications.
Evaluating Results
While VNC reconstruction scored 63% of patients with CAC TNC correctly, PC scored 54% of patients accurately. VNC reconstructions were able to detect the presence of coronary calcium in 90% of patients with CAC TNC, while PC reconstructions were able to detect it in 92% of these patients.
VNC correctly classified 45% of patients according to risk categories, while PC correctly classified 68% of patients.
From these findings, van der Bie drew a number of conclusions.
“First, with photon-counting CT it is feasible to remove the signal of the contrast agents accurately from contrast-enhanced scans,” van der Bie said. “Both reconstructions can more accurately detect the presence of coronary calcium than the absence of coronary calcification.”
In general, the investigators observed that the VNC algorithm underestimates and the PC algorithm overestimates CAC scores compared to TNC scans.
“Most importantly, novel PC reconstructions outperform VNC reconstructions regarding the accuracy of CAC quantification and risk classification,” van der Bie said.
“In our results,” she continued, “we observed an overestimation of the pure calcium algorithm, but only in the lower-risk groups. With the further optimization of this algorithm, we believe that PC will be able to replace contrast-enhanced scans in the near future.”
Access the presentation, “Coronary Calcium Scoring on Virtual Non-contrast and Pure Calcium Reconstructions Compared to True Non-contrast Images With Photon-Counting Computed Tomography,” (S2-STCE-2) on demand at Meeting.RSNA.org.
