Ultrasonic Imaging Based Fetal Cardiac Chambers Segmentation Using Discrete Wavelet Transform

摘要

Doppler Ultrasonic imaging has been used to study the structural and functional aspects of fetal cardiac chambers. Especially the quantitative analysis provides the earlier detection of fetal congenital abnormalities. The challenging task for the biomedical research community is to exactly locate the four cardiac chambers and to introduce appropriate segmentation procedure for chamber measurement. This specific research study suggest a semi automated segmentation procedure using discrete wavelet transform for the segmentation of fetal cardiac chambers, Ultrasonic fetal cine loop sequences based on the apical four - chamber view was considered for the simulation study. Given the fetal ultrasonic cine loop sequence, a master frame is derived by applying Horn - Schunck's motion estimation approach to recognize one typical cardiac cycle followed by averaging the frames belonging to one cardiac cycle. To recognize the fetal cardiac chambers (ROI), a semi automated circular ROI of fetal heart was manually selected and discrete Haar Wavelet transform was applied for chamber segmentation. The ratio of left ventricle width to right ventricle width estimation closely matches with the theoretical bound for different gestation week. The proposed semi - automated technique need to be validated with large datasets before being considered for clinical routine.