Acute Lymphoblastic Leukemia Detection Using Depthwise Separable Convolutional Neural Networks
In this paper, we present a neural network with depth wise separable convolutions (Xception) for the identification of leukemic B-lymphoblast cells, commonly known as Acute Lymphocytic Leukemia(ALL). Earliest possible detection of these cancerous cells is required to minimize the physical toll on the patient and the treatment challenges presented by the disease. Through a transfer learning approach, we tested various convolutional neural network algorithms on our augmented microscopic blood smear image dataset to assess the best performing architecture for classifying leukemic cells, resulting in the Xception architecture. We obtained 99% and 91% accuracy on the training and testing sets, respectively. Furthermore, we achieved a recall rate as high as 98%showing good discrimination power against false negatives.
Clinton, Laurence P. Jr.; Somes, Karen M.; Chu, Yongjun; and Javed, Faizan
"Acute Lymphoblastic Leukemia Detection Using Depthwise Separable Convolutional Neural Networks,"
SMU Data Science Review: Vol. 3:
2, Article 4.
Available at: https://scholar.smu.edu/datasciencereview/vol3/iss2/4
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License