Small calcified coronary atherosclerotic plaque simulation model: minimal size and attenuation detectable by 64-MDCT and MicroCT

Zero calcium score may not reflect the absence of calcifications as small calcifications could be missed. This study aimed to evaluate minimal size and minimal attenuation of coronary calcifications detectable by computed tomography (CT) and to determine the minimal spatial resolution required for detecting calcification onset. Using open source CT simulation software, CTSim^?, several 50%-stenotic coronary artery phantoms were designed with 5?μm resolution, realistic morphology and tissue-specific Hounsfield Unit (HU) values. The plaque had an attenuation resembling fibrous plaque and contained a single calcification. X-ray projections were simulated with settings resembling non-contrast-enhanced 64 multi detector-row CT (64-MDCT). Scanning and reconstruction were simulated with spatial resolution of a 64-MDCT (0.4mm) and of a MicroCT (48?μm). Starting from a single calcium granule, the calcification was simulated to grow in size and attenuation until it could be detected using clinically accepted calcium determination scheme on MicroCT and 64-MDCT images. The smallest coronary calcifications detectable at MicroCT and 64-MDCT, which had a realistic attenuation (?1,024 to 3,072?HU), were of 25?μm and 215?μm diameter, respectively. The area was overestimated 7.7 and 8.8 times, respectively. Calcifications with smaller size need to have an unrealistically high attenuation to be detectable by 64-MDCT. In conclusion, 64-MDCT is only able to detect coronary calcifications with minimal diameter of 215?μm. Consequently, early onset of calcification in coronary plaque will remain invisible when using CT and a zero calcium score can not exclude the presence of coronary calcification.     

Non-calcified coronary atherosclerotic plaque visualization on CT: effects of contrast-enhancement and lipid-content fractions

Computed tomography (CT) may characterize lipid-rich and presumably rupture-prone non-calcified coronary atherosclerotic plaque based on its Hounsfield-Unit (HU), but still inconclusively. This study aimed to evaluate factors influencing the HU-value of non-calcified plaque using software simulation. Several realistic virtual plaqueburdened coronary phantoms were constructed at 5 μm resolution. CT scanning was simulated with settings resembling a 64-row multi-detector CT (64-MDCT) and reconstructed at 64-MDCT (0.4 mm) and MicroCT (48 μm) resolutions. Influences of lumen contrast-enhancement, stenosis-grades, and plaque compositions on plaque visualization were analyzed. Lumen contrast-enhancement and mean plaque HU-value were positively correlated (R² > 0.92), with approximately the same slopes for all plaque compositions. Percentage lipid-content and mean plaque HU-value were negatively correlated (R² > 0.98). Stenosis-grade and noise had minimal influence on the correlations. Influence of lumen contrast-enhancement on plaque HU-value was following a specific exponentially declining pattern (y = Ae^?λx + c) from the lumen border until 2-pixel radius. Outside 2-pixel radius, plaque HU-values deviated maximally 5 HU from non-contrast-enhanced reference. Thus, to avoid lumen contrast-enhancement influence, plaques should be measured outside 2-pixel radius from the lumen border. Based on the patterns found, a lumen influence correction algorithm may be developed. HU-based plaque percentage lipid-content determination might serve as an alternative plaque characterization method. However, its applicability is still hindered by many inherent limitations.     

Assessment of image quality of 64-row Dual Source versus Single Source CT coronary angiography on heart rate: A phantom study

Purpose

To assess the influence of temporal resolution on image quality of computed tomographic (CT) coronary angiography by comparing 64-row Dual Source CT (DSCT) and Single Source CT (SSCT) at different heart rates.

Methods

An anthropomorphic moving heart phantom was scanned at rest, and at 50 beats per minute (bpm) up to 110 bpm, with intervals of 10 bpm. 3D volume rendered images and curved multi-planar reconstructions (MPRs) were acquired and image quality of the coronary arteries was rated on a 5-points scale (1 = poor image quality with many artefacts, 5 = excellent image quality) for each heart rate and each scanner by 3 observers. Paired sample t-test and Wilcoxon Signed Ranks test were used to assess clinically relevant differences between both modalities.

Results

The mean image quality scores at 70, 100 and 110 bpm were significantly higher for DSCT compared to SSCT. The overall mean image quality scores for DSCT (4.2 ± 0.6) and SSCT (3.0 ± 1.1) also differed significantly (p < 0.001).

Conclusion

These initial results show a clinically relevant overall higher image quality for DSCT compared to SSCT, especially at heart rates of 70, 100 and 110 bpm. With its comparatively high image quality and low radiation dose, DSCT appears to be the method of choice in CT coronary angiography at heart rates above 70 bpm.     

Design and Implementation of Gradient Vector Flow Snake to Detect a Reference Object in Pelvic X-Rays for Preoperative Total Hip Arthroplasty Planning Application

The main interest of this research project is to promote automation in performing preoperative planning for hip joint replacement surgery using a special medical image viewing software, ViewPro™. Preoperative planning is performed to carefully prepare the surgery and to accurately select the hip implants. The first step of preoperative planning is to calibrate the x-ray image to adjust the magnification factor. A femoral head implant is used as magnification factor reference. Automation is introduced by developing an algorithm to automatically detect this reference object in the image. The algorithm used for performing the automatic detection of the reference object is gradient vector flow (GVF) snake. A study has been performed to compare the newly developed semiautomatic algorithm to the old manual calibration algorithm. The results show a close relation between the two algorithms (less than 1% of average relative difference). It is concluded that the developed semiautomated algorithm can be used as an alternative for performing the manual calibration.