Analytic Approach for Fetal Head Biometric Measurements Based on Log Gabor Features

This paper presents a new analytical method for fetal biometric measurements. The proposed approach is used to validate doctor’s biometric measurements to detect the malformation of the head. Mainly, biparietal diameter, occipito-frontal diameter, and head circumference are the well-known clinical measurements used to detect the fetal head. Here, the ultrasound images (US) of the gestational age (GA = 22 weeks) are employed using a discriminative set of features. The proposed system consists of the following steps: First, the ultrasound fetal image is passed through the Contourlet preprocessing algorithm to reduce the speckle noise. The resulting image falls to hysteresis thresholding process to preserve the dominant head features. Then, log Gabor features are extracted from the US image to determine the biometric head measurements. The obtained measurements are compared to Expert results using normal and abnormal data sets. The proposed method has been tested on 50 normal and 10 neurological abnormal fetal head US sequences and it reach 98% of accuracy. The main advantages of the new proposed biometric measurement technique versus the existing procedures are: (1) automatic measure of the fetal head displayed on the computer screen of the expert machine without manual intervention; (2) the approach should be implemented and the biometric measure can be extracted automatically from the clinical US image.