Post-processing calcium subtraction method to minimize stenosis-overestimation by blooming artifact

CT images are often affected by blooming artifacts during the diagnosis that facilitate an overestimation of the expression of calcification stenosis, thereby impeding the accurate diagnosis of this condition.

Arterial calcification can act as a blooming artifact in computed tomography (CT) images, leading to overestimations of the blood vessel and the size of calcified plaque. This study proposes an improved CT post-processing method that accurately measures calcium and lumen size in blood vessels.

Six hundred and thirty calcium datasets were obtained from 63 patients diagnosed with a vascular disease. Patients were grouped into three sets corresponding to each image acquisition method used: G1, for the invasive coronary angiography (ICA); G2, for multiplanar reconstruction (MPR) imaging and post-processing; and G3, for the novel method of mixed Gaussian filter and K-mean clustering (GK). Results of GK were generated by adding Gaussian and k-mean clustering algorithms to the MPR post-processing procedure. The analysis of variance (ANOVA), linear regression, and intraclass correlation coefficient (ICC) were used to compare the accuracy and sensitivity of the different methods. All measurements were performed multiple times to mitigate human error.

The ANOVA test revealed no significant differences between the G1 and G3 groups. Hence, linear regression was used to analyze the correlation between the G1 and G3 groups (p< 0.05, R2 = 0.885), and a higher correlation than G1 and G2 was reported (p> 0.05, R2 = 0.432). ICC was performed for reproducibility, wherein high correlation was identified among all groups.

Results of the study indicate that the GK method yields images that are very similar to ICA image measurements. This suggests that the GK can be used as a more effective post-processing method over the inaccurate MPR while remaining non-intrusive when determining the arterial stenosis degree, unlike the ICA.