Pattern Recognition, Imaging Models, and Artificial Intelligence

Abstract

Machine learning and deep learning techniques are always compressed, convergence models. The most appropriate optimization would be the gradient descent, which is capable of finding the best solution with minimal cost. This algorithm can also be used with convolution neural network for better performance. Here a research work proposes a novel automatic image segmentation technique to discriminate between normal and acute lymphoblastic leukemia on bone marrow samples and has been done with the proposed GD_CNN. This method consists of background extraction and RBC separation. In the next stage, the median filtering and morphology technique is used to remove the noise and outliers. An improved GD_CNN model is used to classify normal or abnormal bone marrow samples. This proposed work is implemented using MATLAB 10.0 tool and tested on the bone marrow samples taken from ALL-IDB (Acute Lymphoblastic Leukemia Image Database for image processing) database. The proposed model provides high-quality classification accuracy to identify acute lymphoblastic leukemia.