Application of Machine Learning to Distinguish Premature Leukemia Cells from Healthy Blood Cells

Convolutional Neural Networks (CNNS) to solve this problem have been created with classification validation accuracy rates as high as 96.15%.

Leukemia is a disease that primarily impacts children. The likelihood of beating this deadly disease increases with early detection. Because premature Leukemia blasts are unidentifiable by the human eye, different methods are needed to accurately diagnose cancer in its early stages. Convolutional Neural Networks (CNNS) to solve this problem have been created with classification validation accuracy rates as high as 96.15%. A dataset from the Cancer Imaging Archive [1] was used to train a CNN model. Of the 15,135 images in the dataset, 600 were used to prevent system crashing. With the limited resources available, a 63.6% validation accuracy was achieved.

Explore More!

Research Paper