Band classification based on chromosome shapes

For an automatic chromosome classification, band resolution information is required in order to perform diagnosis on numerical and structural abnormalities. Metaphases with low band resolution are normal used for numerical abnormality analysis while metaphases with high band resolution are used for structural abnormality analysis. In our work, we classify metaphases into low and high band resolution groups using chromosome shape. This band classification task can automatically perform without human intervention leading to faster the diagnosis process. The results showed that chromosome shape information is able to classify metaphases into low and high band resolution groups with the accuracy of 73.08% and 95.24%, respectively.