Abstract
Coronary computed tomography angiography (CCTA) is the most commonly used image acquisition method for the diagnosis of cardiovascular disease (CVD) currently. While plaque recognition and classification based on CCTA image sequences are the most common diagnostic methods in clinical practice. This paper proposes a plaque detection and recognition algorithm based on CCTA image sequences, which using frame-gradient of Gaussian variance with MIMICS and MATLAB software. This algorithm is different from the traditional algorithms based on single frame image processing and recognition. It innovatively uses the variance gradient between multiple frames as a plaque detection method, and smooth the variance gradient noise with one-dimensional Gaussian filter. For further classifying plaques, we perform threshold detection with multi- feature value including maximum and minimum, and optimize the threshold for detection, after that we remove noise by continuous point filter(CPF). Finally, the detection formula is optimized, and the result of multi vessel detection is displayed with amplitude overlay chart(AOC). By verifying 80 CCTA image sequences, the experimental results show that the total accuracy rate up to 83.75%, which indicate the high application value.