| Heart is one of most important parts in human body, and its functional status determines thehuman’s healthy. As the pump of blood cycle, left ventricle (LV) pumps the blood to circulatethe whole body with a series of systolic and diastolic motion. Therefore, analysis of LV motionis helpful to the diagnosis of the heart disease. Modeling of LV3D model based on medicalimages and then analyzing LV motion, valuing LV function using the3D model are the popularresearches in recent years. And it is primary to analyze dynamic parameters which can reflect thelocal LV functional capability to make diagnosis and treatment in earlier stage of cardiac disease.LV contour in CT image is not a regular ellipse and not close, because the left atrium (LA)covers a part of LV during scanning, so modeling based on searching grayscale and matching isnot suitable to cardiac CT images. There is not so much literature on cardiac motion with CTimages, which is also lack of relative research on statistical analysis of dynamic parameters. Thispaper presents a kind of LV point distribution model (PDM) based on CT images, distractsdynamic parameters, and analyzes statistically dynamic parameters to judge the function of LV.The main work and contribution of this paper are as follow1. It proposes a new method to reconstruct continuous LV PDM based on cardiac CTimages. It contains of manual segmentation, point cloud obtaining, NURBS surface fitting,re-sampling and point cloud matching. Manual segmentation and NURBS fitting are optimizedbased on the characters of the heart and the data to make the model data more accurate andsurface more smooth. Finally continuous LV PDMs are reconstructed in different physical phase.2. Torsion angle and torsion angle velocity are distracted from the LV PDM. And alsocardiac velocity, acceleration, torsion angle, torsion angle velocity are calculated in differentphysical phase based on LV PDM.3. The situation of abnormal motion is figured out towards the analysis of parameters. Andalso the part of the abnormal motion is classed out towards the clustering algorithm. Withstatistical analysis, it’s helpful to diagnose heart disease objectively in real time. |
