Sickle cell anemia is a red blood cell disorder that causes a shortage of oxy- gen due to the sickling of red blood cells in the body. Sickle cell disease is a genetic disorder and typically remains undetected during a human life- time. We have developed a novel image processing-based system that can detect sickle cell anemia from brightfield microscopic images. An input image is passed through three pipelines that perform a variety of distance image processing operations such as top-hat filtering, local adaptive thresholding, morphological erosion, Canny edge detection, contour extraction, etc. The outputs of the three pipelines are then combined using an OR operation and passed through an area-based filtering operation to identify sickle cells and healthy cells. We have trained and evaluated our model on a dataset sourced from Kaggle, public datasets, and images of RBCs collected by us. We were able to achieve accuracy and precision of 96.46 % and 94.82 %, respectively, on the dataset. We could classify sickle cell anemia disease as well as trait patients. We expect that our proposed algorithm can be used in remote locations where a fast diagnosis is needed in resource-limited settings.

microscopic images, sickle cell anemia, image processing