Classification of Chromosomes to Diagnose Chromosomal Abnormalities Using CNN

Karyotyping is a crucial screening test for genetic illnesses. However, as it is currently practiced, it requires considerable manual effort and experience to interpret results, which renders it time-consuming and costly. Automating karyotyping is important and popular, as it allows for less time-intensive procedures. It can also be used to help classify chromosomes into 23 types. ConvNet (Convolutional NN) based deep structured learning was employed to classify 10304 chromosome images, with trained and tested datasets containing 4830 chromosomes. The method we propose is accurate 92.5 percent of the time, the approach outlined by the study's authors fared better than other approaches reported in the literature. We created a statistic called “percentage of well classified karyotypes” to examine how useful our technology is to medical professionals. When tested on this metric, we achieved an accuracy of 91.3 percent, indicating that our modelwill be used by doctors are in genetic diagnosis.