Acute lymphocytic leukemia detection using hybrid deep learning models

Abstract

Acute Lymphocytic Leukemia is a type of cancer that affects white blood cells and it spreads quickly. This study proposes a computer-aided diagnosis system to detect this type of leukemia from blood microscopic images. We introduce a hybrid machine learning model that uses a ResNet18 encoder to extract latent embeddings from the multi-otsu segmented white blood cells and we feed those embeddings into machine learning classifiers. The random forest and the k-nearest neighbours recorded the best classification accuracy i.e. 98% while misclassifying two samples from the ALL-IDB dataset.