부산대학교 도서관

Acute lymphoblastic leukemia is a condition that occurs in the bone marrow. Most cases of Acute Lymphoblastic Leukemia (ALL) develop from cells that would turn into white blood cells (other than lymphocytes), but some cases of ALL develop in blood-forming cells of various kinds. Anything begins in the bone marrow (the soft inner part of some bones, where new blood cells are made), yet it moves rapidly into the blood by and large. It may spread to various parts of the body once in a while, including the lymph nodes, liver, spleen, central nervous system (cerebrum and spinal cord), and testicles. The traditional method used by doctors to detect ALL is tedious and time consuming. It consists of several methods such as blood testing, aspiration and biopsy to the bone marrow, image testing using CT scanning and MRI. These methods take time and are expensive. The proposed approach uses imaging processing techniques to detect irregular blood cells in the microscopic images of the blood. Using the training data collection, the program is trained using DCNN and is uploaded into HDFS Hadoop framework. The learned method classifies when a new image is inserted into the program whether the collected picture includes cancerous leukemia cells or irregular cells. This way the process of detection of leukemia is automated and can be accessed by any of the hospitals with connected to the cloud system.