Identification of Abnormal Red Blood Cells and Diagnosing Specific Types of Anemia Using Image Processing and Support Vector Machine

Blood is an important liquid that supports performs important functions for human. The general components of human blood are RBCs, WBCs, plasma, and platelets. Whereas, RBCs occupies the largest volume in blood. RBCs can be healthy or unhealthy. An unhealthy or abnormal red blood cells indicates blood related disorders, such as Anemia. RBCs can be determined by variations in size, shape, and color. The method in evaluating RBC is through conventional microscopy which leads to misreport results and experts are given hefty workload. A proposed system is aimed to support medical technicians, hematologist, and pathologist in identifying RBC by creating a device using Raspberry Pi that can measure the different parameter of the RBC such as area, perimeter, diameter, shape geometric factor (SGF), and detecting the central pallor and target flag. Related studies were already conducted using different approach one study uses Artificial Neural Network (ANN) in classifying RBC with an accuracy 90.54%. Another study uses radial basis function network it obtained an accuracy of 83.3%. With the proposed system using Support Vector Machine (SVM) classifier it acquired an accuracy of 93.33% in identifying 7 different red blood cells such as normal, echinocytes, elliptocytes, dacrocytes, spherocytes, target cell, stomatocytes, and identifying unknown cells. The classification of RBC helps in diagnosing different types of anemia such as iron-deficiency anemia, thalassemia, hereditary spherocytosis, and myelophthisic anemia. This system serves only as an aid for doctors for early diagnosis of identifying abnormal red blood cells and succeeding laboratory exams must be done to finally conclude a disease associated to abnormal RBCs.