Content Based Retrival Of Malaria Positive Images From A Clinical Database

Modern hospitals are attempting to create a database of patients' diagnostic history that also contains multiple images taken during different clinical tests. This has lead to a demand for easy retrieval of images matching a query condition so that this database can be used as a clinical decision support system. This paper presents an efficient technique for retrieval of malaria positive images from a clinical image database, that match a query condition mentioned in text or as another malaria positive image. The proposed method identifies the presence of malaria by identifying the presence of chromatin dots within RBCs. The algorithm utilizes 4-connected labeled region maps to analyze and modify the image, i.e., delete unwanted artifacts, etc. These are also used to count RBCs. Alternative methods for identifying chromatin dots within the RBCs are also presented. Precision and recall efficiency of the algorithm was also studied.