A Novel Approach for Objective Assessment of White Blood Cells Using Computational Vision Algorithms

Table 4

Comparison between systems semiautomatic versus developed algorithm.

Author	Techniques	Results
[7]	Segmentation by OTSU and classification by neural networks.	Average performance of 65% and 95% after training.
[12]	Algorithm based on gram-Schmidt orthogonalization.	Average performance of 85.4%.
[11]	Iterative method of increasing region.	Effectiveness was obtained to identify the cells of 76.47% for Basophils, 95.5% for neutrophils.
[17]	Contrast adjustment in RGB and complex-value neural networks.	To precision in complex value of 99.3% and 97.5% in real value was obtained.
[18]	Image segmentation with contrast adjustment and filtering in grayscale.	An accuracy of 80.04%, 69.3%, 86.3%, 80.3% and 83.8% was obtained for basophils, eosinophils, monocytes, neutrophils and lymphocytes.
[19]	Classification by PCA and Dendrodendritic.	The average efficiency of the process was 77.2%.
[20]	The overlapped Detection of red blood cells in microscopic images of blood smear.	Sensitivity and specificity percentages were obtained higher than 96%
[21]	Classification of different types of white blood cells by global threshold and features geometrics.	Percentages of classification were obtained higher than 98%, 92% and 95% for lymphocyte, monocyte and neutrophil respectively.
[22]	Leukocyte nucleus segmentation and recognition by K-Means clustering.	Was obtained to precision of 98% for Basophil, 98% Eosinophil, 84.3% 93.3% Lymphocyte, monocyte and neutrophil 81.3.
[23]	Leukocytes Classification In Blood Smear by support vector machines (SVM).	Was obtained to accuracy of 98.5%, 99.9% Neutrophil for Eosinophil, 98.8% 93.7% Lymphocyte and Monocyte.
[16]	WBC Segmentation and Classification by Fuzzy C-Mean.	The accuracy of the process was 91% for the 5 types of cells.
Proposed System	Classification of cells by networks of Gaussian radial basis functions (RBFN) and morphological descriptors.	Was obtained to 100% accuracy of 73.07%, 93.42%, 97.37% and 79.52% for Basophiles, Eosinophil’s, lymphocytes, monocytes and neutrophils respectively and 98.2%.