Computerized automatic nasal bone detection based on ultrasound fetal images using cross correlation techniques

Abstract

Recent study shows that fetal chromosomal abnormalities can be detected in early ultrasonic prenatal screening by identifying the absence of fetus nasal bone. The drawbacks of current method are operator dependent, observer variability and improper training. In particular, accurate nasal bone detection requires highly trained sonographers, obstetricians and fetal medicine specialists since the ultrasound markers may easily confuse with noise and echogenic line in ultrasound image background. We present a computerized method of detecting the absence of nasal bone by using normalized grayscale cross correlation techniques. Image preprocessing is implemented prior to cross correlation to assess the availability of nasal bone. The resultant threshold, bordering the absence and presence of nasal bone, is set to a value of 0.35. The accuracy of the developed algorithm achieved was, around 96.26 percent which promises an efficient method to recognize nasal bone automatically. The threshold can be further improved if a larger set of nasal bone ultrasound images are applied.