Quick Leukocyte Nucleus Segmentation in Leukocyte Counting

Application of image recognition technology in pathological diagnosis of blood smears

Traditional manual blood smear diagnosis methods are time-consuming and prone to errors, often relying heavily on the experience of clinical laboratory analysts for accuracy. As breakthroughs in key technologies such as neural networks and deep learning continue to drive digital transformation in the medical field, image recognition technology is increasingly being leveraged to enhance existing medical processes. In recent years, advancements in computer technology have led to improved efficiency in the identification of blood cells in blood smears through the use of image recognition technology. This paper provides a comprehensive summary of the methods and steps involved in utilizing image recognition algorithms for diagnosing diseases in blood smears, with a focus on malaria and leukemia. Furthermore, it offers a forward-looking research direction for the development of a comprehensive blood cell pathological detection system.

Leukocyte classification based on statistical measures of radon transform for monitoring health condition

In the medical field, automated and computerised analytic tools are essential for faster disease diagnosis. The main objective of this research work is to classify the leukocytes accurately into four different subtypes based on the pattern of the nucleus. The features are extracted from the segmented nucleus, which play a vital role in the pattern recognition. The technique comprises a novel idea of computing the statistical measures such as peak difference and standard deviation of the radon transformed graph for a single angle of rotation along with other features. Three Gray Level Co-occurrence Matrix (GLCM) based features, two geometric features and four RST moment invariants are also extracted for feature fusion. The fused feature vectors are trained and evaluated using random forest classification algorithm.This method provides an overall accuracy of 97.61% and it is able to determine the lymphocyte, neutrophil and eosinophil with 100% accuracy. The classification without incorporating radon transform features is also performed which provides an accuracy of only 80.95%.

A Polylobar Nucleus Identifying and Extracting Method for Leukocyte Counting

Accurate counting of leukocytes is an important method for diagnosing human blood diseases. Because most nuclei of neutrophils and eosinophils are polylobar, it is easily confused with the unilobar nuclei in nucleus segmentation. Therefore, it is very essential to accurately identify and determine the polylobar leukocytes. In this paper, a polylobar nucleus identification and extracting method is proposed. Firstly, by using the Otsu threshold and area threshold method, the nuclei of leukocytes are accurately segmented. According to the morphological characteristics of polylobar leukocytes, the edges of the mitotic polylobar leukocytes are detected, and the numbers of polylobar leukocytes are determined according to the minimal distance rule. Therefore, the accurate counting of leukocytes can be realized. From the experimental results, we can see that using the Otsu method and the area threshold to segment the polylobar nuclear leukocytes, the segmentation ratio of the leukocyte nucleus reached 98.3%. After using the morphological features, the polylobar nuclear leukocytes can be accurately counted. The experimental results have verified the feasibility and practicability of the proposed method.