Supervised chromosomal anomaly detection using VGG-16 CNN model

Feature Fusion Classifier With Dynamic Weights for Abnormality Detection of Amniotic Fluid Cell Chromosome

“…Compared with He [23], which detected trisomy 21 with an accuracy of only 85.2%, and Mona [24], which only detected chromosome number abnormalities, the detection scope of this study is wider, and the results are stable, and it is not limited to the detection of single chromosome abnormalities or only the detection of chromosome number abnormalities. Nimitha [26] and Ezhumalai [27] similarly, only abnormal chromosome number was detected, but abnormal chromosome deconstruction was not detected.…”

“…Yang et al [25] utilized the deep convolutional neural network (DCNN) model to classify 2424 normal chromosomes and 544 abnormal chromosomes were classified, including 24 normal chromosomes (autosomal 1-22 and sex X or Y) and 8 abnormal chromosomes, with a classification accuracy of 87.76%. Nimitha et al [26] used VGG16 for transfer learning, and the classification accuracy of chromosome number abnormalities was 95.5%. Ezhumalai et al [27] used the DCNN model to distinguish five chromosome number abnormalities, Trisomy 13 syndrome, trisomy 18 syndrome, trisomy 21 syndrome, trisomy XXY syndrome, and X chromosome.…”