Based On The Echocardiographic Heart Function Index Of Computer Measurement

The cardiac function is an important physiological index,which is necessary for the ultrasound examination.At present,the EF index is often adopted to evaluate the cardiac function.Doctors obtain images by the ultrasound,then measured some parameters manually,and finally fill results into the formula to calculate the index.A single index is not a comprehensive evaluation of the cardiac function.To solve this problem,adding E/A value and the TEI index can be a more comprehensive.For the purpose of solving the defect of the manual measurement,we use image processing methods to complete the measurement and calculation automatically.The measurement process is simplified,and the workload is reduced.Further,if the new design is applied to the hospital PACS system and the ultrasound reporting system,it will play a better role.The original image is captured from the ultrasound.We pick up the region of interest and do some preprocessing.The pre-processing includes:a two-dimensional median filter to smooth the image;the non-linear gray-scale transformation to increase the dynamic range of the image;the morphology function to smooth the border and remove the background interference.After the pre-processing,we obtain the initial envelope from the binary image,calculate the correlation-function of the envelope,and use the method of height-segmentation and maxima-accumulation method to obtain the cardiac cycle.A single-cycle image is then picked out.The accurate envelope of the blood flow spectrum both at mitral and aortic valve are obtained by the snake model based on the dynamic programming method.By using the Graham scanning for the convex hull extraction and fitting these convex hull points with the conics,roots of the conics are regarded as points we need.The method of the border extraction based on the three-dimensional scanning is adopted to deal with the M model ultrasound at the mitral.The maximum and minimum values can be calculated between two borders.By filling all useful values into the formula,we can obtain the cardiac function index.To facilitate the testing,we produced an interactive interface program.Mitral and aortic valve Doppler ultrasound images from 30 patients are acquired by Philips IE33 ultrasound apparatus.These images are prepared for the E/A and TEI index test.M-model images from 30 patients are also acquired for the EF index test. All calculation results are compared with those from the manual measurement.The mean absolute errors are calculated with the golden standard of the manual measurement.The mean absolute error of E/A,TEI and EF is 4.3%,10.7%and 10.4% respectively.Three parameters which are the heart rate,LVDd and LVDs are also obtained at the same time.Results show that this new system of the cardiac function measurement has a better accuracy.By using the computer image processing,the complexity is reduced, while the efficiency improved.Therefore,this new system may have a certain clinical applications in the future.