No abstract

Abstract. In white blood cell (WBC) diagnosis, the most crucial measurement parameter is the WBC counting. Such information is widely used to evaluate the effectiveness of cancer therapy and to diagnose several hidden infection within human body. The current practice of manual WBC counting is laborious and a very subjective assessment which leads to the invention of computer aided system (CAS) with rigorous image processing solution. In the CAS counting work, segmentation is the crucial step to ensure the accuracy of the counted cell. The optimal segmentation strategy that can work under various blood smeared image acquisition conditions is remain a great challenge. In this paper, a comparison between different segmentation methods based on color space analysis to get the best counting outcome is elaborated. Initially, color space correction is applied to the original blood smeared image to standardize the image color intensity level. Next, white blood cell segmentation is performed by using combination of several color analysis subtraction which are RGB, CMYK and HSV, and Otsu thresholding. Noises and unwanted regions that present after the segmentation process is eliminated by applying a combination of morphological and Connected Component Labelling (CCL) filter. Eventually, Circle Hough Transform (CHT) method is applied to the segmented image to estimate the number of WBC including the one under the clump region. From the experiment, it is found that G-S yields the best performance.

<span>Blood cells diagnosis is becoming essential to ensure a proper treatment can be proposed to a blood related disease patient. In current research trending, automated complete blood count analysis system is required for pathologists or researchers to count the blood cells from the blood smeared images. Hence, a portable mobile-based complete blood count (CBC) analysis framework with the aid of microscope is proposed, and the smartphone camera is mounted to the viewing port of the light microscope by adding a smartphone support. Initially, the blood smeared image is acquired from a light microscope with objective zoom of 100X magnifications view the eyepiece zoom of 10X magnification, then captured by the smartphone camera. Next, the areas constitute to the WBC and RBC are extracted using combination of color space analysis, threshold and Otsu procedure. Then, the number of corresponding cells are counted using topological structural analysis, and the cells in clumped region is estimated using Hough Circle Transform (HCT) procedure. After that, the analysis results are saved in the database, and shown in the user interface of the smartphone application. Experimental results show the developed system can gain 92.93% accuracy for counting the RBC whereas 100% for counting the WBC.</span>

White Blood Cells (WBCs) analysis is an important procedure to detect diseases is that closely related to human immunity system. Manual WBCs analysis is laborious and hence computer aided system (CAD) is a better option to alleviate the shortcoming. Since conventional segmentation-classification approach is tedious to configure, a Convolutional Neural Network (CNN) become recent trend for WBCs classification. Previously, there are many works proposed for WBCs identification. However, the models that can be generalised to works well among various datasets is remain vague. In this paper, an analysis of various CNN models which are simple Alexnet, embedded friendly Mobilenet, inception based Googlenet, systematic architecture VGG-16 and skip connection based model (Resnet & Densenet), are tested with three major WBCs datasets (Kaggle, LISC and IDB-2). From the rigorous experiments, it can be concluded that simple CNN model of Alexnet performs well across all three datasets with 98.08% accuracy on Kaggle, 96.34% accuracy on IDB-2 and 84.52% on LISC. This outcome can be utilise as a basis to improve the CNN classification model that can be generalize to works under various WBCs datasets.

No abstract