Performance analysis of Gaussian and bilateral filter in case of determination the fetal length

Abstract

One step in the image processing is filtering that located in the preprocessing. In the context of fetal analysis on the ultrasound image, filtering is really needed to enhance the quality of ultrasound image. This study conducted analysis of performance between Gaussian and bilateral filter in the fetal length. Peak signal to noise ratio (PSNR) was used to measure the quality of reconstruction the image compression. In case of early pregnancy, fetal analysis is important to determine some disabilities that might cause miscarriage. One indicator parameter for determine the fetal health on the early pregnancy is Crown Rump Length (CRL) or fetal length. Calculation of fetal length was performed with different filter (Gaussian and bilateral) after segmentation carried out. According to the experimental data, using Gaussian filter with kernel 3×3; 5×5; and 7×7, PSNR was achieved consecutively 31 dB, 28 dB, and 27dB. Furthermore, in the same ultrasound image, PSNR is 30 dB when the filter is bilateral. At the same time, fetal length was also counted separately for Gaussian and bilateral filter. The results of mean fetal length were 6.5 cm, 9.9 cm, 9.4 cm for Gaussian filter with kernel: 3×3, 5×5, 7×7 respectively; while the system was obtained fetal length 5.6 cm for bilateral filter. It is seen that by using a bilateral filter, the fetal length achieved quite accurately characterized with a mean error of fetal length is quite small.