Acute Lymphoblastic Leukemia Detection Using Sequential, Lenet and Vggnet Models

Main Article Content

Bhuvaneswari S, Abirami.A, Kavitha Datchanamoorthy, Haritha M, Archita A S, Kavya Varshini M

Abstract

The purpose of this research is to develop a web-based application for blood cancer cell detection using deep learning to provide accurate identification of acute lymphoblastic leukemia (ALL) and its subtypes. A conclusive diagnosis of acute lymphoblastic leukemia (ALL), as the most prevalent cancer, necessitates intrusion, expensive, and diagnostic test takes a long time. It is a tedious process to analyse acute lymphoblastic leukemia if the number of patients is quite high. In order to prevent mortality, it is important to identify acute lymphoblastic leukemia at an early stage. In the primary separating cancer from non-cancer instances, images of peripheral blood smear using ALL diagnosis plays an important role. Laboratory users' examination of these PBS pictures is plagued with issues such as diagnostic mistakes, as the wrong status of ALL symptoms frequently leads to an incorrect diagnosis. Taleqani Hospital (Tehran, Iran) takes the responsibility of preparation of images of this dataset. It is handled by their bone marrow laboratory. The blood samples from 89 patients are taken from peripheral blood smear, where 64 patients show the symptoms of ALL and rest of the people are healthy. The total image used in this dataset is 3256.All photographs were captured using and saved in JPG format. Haematologist find it helpful to identify the accurate condition of the patient in a quick and easy way. There are two classes namely benign and malignant. Pro-B, Early Pre-B and Pre-B cells are subtypes of Malignant. In this project, microscopic blood cell images are taken as input. The models used are Sequential Model, VGGNet and LeNet. Based on the comparison between these architectures, a conclusion has been arrived that LeNet is the most suitable architecture as it produces an accuracy of 95.75% and takes less computational time.

Article Details

Section
Articles