Automated Measurement Method Of Common Carotid Arteryintima-media Thickness In Ultrasound Image Based On Markov Random Field Models

Common carotid artery intima-media thickness(IMT)is regarded as one of the most important indicators for the assessment of cardiovascular diseases.In clinical practice,IMT can be measured by manual tracing.In order to solve the problems such as laborious works and highly user dependent,and to meet the requirements of the automated and efficient measurement,this paper design an automatic,fast and accurate approach for IMT measurements.This thesis proposes an automated method based on Markov random field models for IMT measurement.The label field is initialized from line segments generated by Hough transform and the iterated conditional modes(ICM)method is applied to detect the final boundaries of the lumen-intima interface(LII),as well as the media-adventitia interface(MAI).The approach can be divided into several cascaded steps:(1)The ROI extraction,(2)Bilateral Filtering,(3)Edge Map,(4)Hough Transform of the sub-image,(5)The Initial Label Field,(6)Final Boundaries,(7)Post-processing stage.A total of 80 ultrasound images from 80 corresponding patients were tested with the proposed method.The ground truth(GT)of the IMT was manually measured by two experts and then averaged,and one expert completed four times measurements in total,while the automatic segmented(AS)IMT was computed by the proposed method.The mean of the absolute error±standard deviation between AS and GT IMT was 0.0244±0.0227 mm,and the correlation coefficient between the two is 0.9810.Furthermore,the computational time was 1.7 seconds per image by serial computation with MATLAB under Windows7-32 bit,while using Intel Corei5-4200 M and 4GB RAM.Experimental results proved that the proposed method is not only efficient yet also accurate.