MCV Measurement of Abnormal Red Blood Cells Using Adaptive Neuro-Fuzzy System with Image Processing

Background & Aims: Red blood cells features like size, shape, and volume are important factors in diagnosing related blood disorders like iron deficiency and anemia. Every day, thousands of blood samples are tested by microscope and cell counter in pathology laboratories around the world which is time consumingly and costly. Therefore an automated system for classifying blood samples to normal and abnormal could be effective in time and cost. Materials & Methods: Adaptive local thresholding and bounding box methods are used to extract inner and outer diameters of red cells. An adaptive network-based fuzzy inference system )ANFIS) is used to classify blood samples to normal and abnormal. Results: Accuracy of the proposed method and area under ROC curve are 96.6% and 0.9950 respectively. Conclusion: This work aims at providing an automated system for classifying blood samples to normal and abnormal based on the microscopic images of red cells. Using local thresholding to find the optimum threshold in each sub-image can compensate brightness variations between sub-images. In removing outliers, by using ±30% window, some of the useful data are lost and the ±70% window does not filter out outliers sufficiently. The results of cell counters depend on environmental conditions, storage conditions of blood samples, anticoagulant, the number of red blood cells that stick together, and white blood cells in the sample. The proposed method enables us to diagnose with just one blood drop and works well on pathological images. It could provide a cost effective alternative to existing routines in laboratories.