A new hybrid framework based on deep neural networks and JAYA optimization algorithm for feature selection using SVM applied to classification of acute lymphoblastic Leukaemia

Abstract

ABSTRACT Identification of Acute lymphoblastic leukemia (ALL) in microscopic images is one of the most challenging tasks in medical image analysis. Despite a wide variety of image processing and deep learning techniques, the task of extracting features from ALL images and selecting the proper features from these redundant and high-dimensional feature space set, and then detecting ALL cells is still a complex issue. In this study, we present a new hybrid two-layer framework to construct an appropriate subset of features. In the first layer of the proposed method, image segmentation is implemented using an improved modified First-Spike-based approach integrated with a Gaussian function. Then a developed deep residual architecture is employed to extract the features. In the second layer, a powerful and reliable meta-heuristic algorithm known as the JAYA optimization algorithm is adopted for feature selection due to its computational efficiency. Finally, a support vector machine (SVM) is used to classify ALL images. To show the effectiveness of the proposed model, it is applied on microscopic images of blood samples from ALL images (ALL-IDB) and ISBI-2019 C-NMC dataset. The results show the superiority of the model to be an appropriate choice for future biomedical imaging tasks.