CIRA (Chromosomal Image Recognition Algorithm) Image Processing and ML
Runs on red cell images.
Example Data:
More examples and results can be viewed in the Image Results folder.
The clustering algorithms were used as classifying algorithms by clustering the data with varying values for K then labeling each cluster with the class that occured the most. When a new cell needed to be classified it was placed in the closest cluster and given the same label as that cluster. Accuracy was calculated using this method and the original class labels.
Accuracy: 0.791
Image 1. K-Means K=2 Graph: This shows the best clustering result among the 3 chosen methods K-Means, with K=2. The left side would be the unhealthy cluster, and the right side would be the healthy cluster.
Accuracy: 0.610
Accuracy: 0.712
Accuracy 0.4975
Accuracy: 0.87
Accuracy: 0.86025