Automated Trisomy 21 Assessment Based on Maternal Serum Markers Using Trivariate Lognormal Distribution

摘要

Trisomy 21 is the most frequent types of chromosomal abnormalities. Generally, current methods for trisomy risk assessment are divided into two techniques, which are invasive and non-invasive methods. Invasive methods are including amniocentesis, chorionic villus sampling (CVS), or percutaneous umbilical cord blood sampling (PUBS), but its drawbacks are expensive, time consuming and having risk of miscarriage, where else non-invasive methods are counting on ultrasound marker and maternal serum markers screening. Nevertheless, single evaluations on ultrasound markers itself are always not enough for risk assessment in terms of its accuracy, reliability and consistency. So, we proposed a new mathematical algorithm which combines three maternal serum markers using trivariate lognormal distribution to calculate automatically the probability or likelihood that a woman has an affected pregnancy or not. The developed algorithm was implemented into graphical user interface to act as computer aided e-diagnostic system. We have compared the results with published finding and found it is almost equally accurate.