A Decision Support System (Dss) For Classification And Retrieval Of Leukemia From Microscopic Blood Cell Images
MetadataShow full item record
Type of WorkText
DepartmentComputer Science and Bioinformatics Program
ProgramMaster of Science
RightsThis item is made available by Morgan State University for personal, educational, and research purposes in accordance with Title 17 of the U.S. Copyright Law. Other uses may require permission from the copyright owner.
Leukemia accounts for almost a third (30%) of all childhood cancers. Out of the four main types, the Acute Lymphocytic Leukemia (ALL) is the most fatal if left untreated due to its rapid spread into the bloodstream and other vital organs. The detection of ALL in its early stages reduces the mortality considerably. Visual microscopic examination of peripheral blood slides still used as the standard leukemia diagnosis technique. Although, it suffers from problems such as subjective interpretations, operator tiredness and efficiency, the morphological analysis can be easily automated and performed directly in blood microscopic images. The development of a fully automated screening system may provide hematologists with significant aid in the effort to detect and classify leukemia cells more effectively and efficiently. To date, only a few research efforts have focused on finding the likelihood of malignancy based on applying some feature extraction and classification schemes. The objective of this research is to provide hematologist with a screening system for ALL detection and recognition that will be able to respond to image based visual queries by displaying relevant microscopic blood images of past cases, along with the associated pathological diagnosis (classification). This work is focusing on the development of such a system based on image segmentation, feature extraction, image classification, similarity matching, and finally performance evaluation in a public image dataset (ALL-IDB) of peripheral blood samples of normal individuals and leukemic patients. The experimental results demonstrate the effectiveness of our system and show the potential of real clinical integration with 96.62 % accuracy level.