Calcified plaque within the coronary artery is a marker of the degree of underlying atheromatous involvement of that vessel, and therefore can be used as a predictive tool for CHD in both asymptomatic and symptomatic patients. To improve the performance and practicality of the coronary calcification detection method, 22 features charactering the coronary calcifications as well as a SVM classifier were studied and their performance in the detection process was evaluated. Based on this, a novel and convenient detection scheme was proposed. The results showed that the best performance of stage 3 (whether a patient contains coronary calcifications or not) was achieved with sensitivity of 96.88%, specificity of 91.30% and accuracy of 94.55%, while in stage 4 (whether a coronary artery or a coronary branch contains calcifications), the highest average sensitivity, average specificity and average accuracy of each patient are 81.50%, 84.72% and 84.09% respectively.