| Objective:To study the ultrasound images of the fetal Nuchal Translucency(NT)in the normal pregnancy(pregnancy 11+0~13+6weeks)in accordance with international measurement standards:1.The feasibility of clinical application of the new technology which is used to locate fetal Nuchal Translucency(NT)and measure its thickness.2.The feasibility to integrate computer vision,pattern recognition and artificial intelligence technology in information science into traditional fetal ultrasound images.Methods:Two kinds of ultrasonic images of the same case(pregnancy 11+0~13+6weeks,head-to-hip diameter 45mm~84mm,)which was made by NT-qualified,experienced ultrasound doctors who used an American GE-E8 ultrasound machine.Two standard fetal NT images of Figure 1(with NT markers)and Figure 2(without NT markers),and manual measurements were collected,and another operator was shown in Figure 2(without NT-tagged images)in ultrasound.The image was automatically located and the fetal Nuchal Translucency(NT)and the thickness measured.The location and measurement results of the two methods were compared.Results:1.Use automatic Location and Measurement of Nuchal Translucency,222cases of samples were successfully positioned and measured;2 cases of samples were output position,but the position was wrong;45 cases of samples indicated that the picture definition was not qualified and could not be automatically positioned.2.Compare the consistency of manual location and automatic location,use Kappa test.And the result was Kappa≈0.975>0.8,which was highly consistent.3.Compared with manual measurement and automatic measurement,there was no statistical significance.(P≈0.058>0.05);4.The performance analysis of the software shows that the sensitivity and specificity of the software are 100%and 75%respectively and the AUC area is 0.995,which indicates that the software has good performance.Conclusion:It is feasible and helps to standardize the market evaluation system for fetal Nuchal Translucency(NT)testing in early pregnancy that the new software which is used to locate NT and measure its thickness at 11+0~13+6gestational weeks(Crown-rump Length was 45mm~84mm).Clinical trials of this new technology have shown that it is feasible to incorporate computer vision,pattern recognition,and artificial intelligence techniques in information science into traditional fetal ultrasound images. |