First Trimester Diagnosis of Trisomy-21 Using Artificial Neural Networks

Abstract

A chromosomal disorder caused by the presence of all or part of an extra 21st chromosome is known as the Down syndrome, or trisomy 21, or trisomy G. In the last fifteen years it has become possible to observe these features by ultrasound examination in the third month of intrauterine life. About 75% of trisomy 21 fetuses have absent nasal bone. In the present work, neural network schemes that have been applied to a large data base of findings from ultrasounds of fetuses, aiming at generating a predictor for the risk of Down syndrome are reported. A number of feed forward neural structures, both of standard multilayer and multi-slab types were tried to find the best for prediction. The database was composed of 23513 cases of fetuses in UK, provided by the Fetal Medicine Foundation in London. For each pregnant woman, 19 parameters were measured or recorded. Out of these, 11 parameters were considered as the most influential at characterizing the risk for this type of chromosomal defect. The best results obtained were with a multi-slab neural structure. In the training set there was a correct classification of the 98.9% cases of trisomy 21 and in the guidance (test) set 100%. The prediction for the totally unknown verification test set was 93.3%.

Artificial neural networks as clinical decision support systems

“…Neocleous and his fellow researchers 21 employ many different ANN models to classify and predict chromosomal abnormalities. The dataset is divided among the Down syndrome, Edwards syndrome, Patau syndrome, and the Turner syndrome groups.…”

“…The dataset contains 31,611 cases and the number of feed‐forward ANN structures, both standard type and multi‐slab are tested with a different number of neurons, layers and activation functions to find the optimal one. The 5 layer ANN topology 10‐80‐10‐80‐5 performance is reported to be the most optimal in case of the Down syndrome with the classification accuracy of 78.9% but for the other three syndromes, the number of examples for each one is less and so the diagnostic results are not very impressive 21 . The approach is non‐invasive for the classification of multiple syndromes using a single neural network.…”

“…However, ANN architecture is very complex causing the training and testing to slow down for the requirement of heavy computational power. In another approach mentioned at, 21 multiple feed‐forward neural structures, both standard multilayer and multi‐slab are used to predict the disease using ultrasound images. The dataset contains 23,513 cases identified with 19 parameters as the most important for distinguishing this type of chromosomal deficiency.…”