Automatic Segmentation and Classification of White Blood Cells in Peripheral Blood Samples

Abstract

White blood cells analysis is generally performed for helping specialists in evaluating a wide range of hematic pathologies such as acquired immune deficiency syndrome (AIDS), blood cancer (leukemia) and other related diseases. Segmentation, Counting, and classification of leukocytes or white blood cells (WBC) in the peripheral blood samples images provide informative data about the samples. Therefore, performing them in the most efficient way is very important in the hematological analysis procedure. Unfortunately, the traditional manual segmentation method is very tedious, time-consuming and provides inaccurate results due to the human factor. Hence, a computer-aided segmentation and classification system is needed to make the process both fast and accurate. In this paper, a new and completely automated system for different types of WBCs segmentation is proposed. The system is evaluated on peripheral blood smears from whole slide images based on color space transformation. In the segmentation process of the WBC, image enhancement techniques were applied on saturation frame of HSV color space and a simple thresholding technique was used to find the brightest object. Moreover, three types of features were extracted from the segmented WBCs which are morphological, statistical, and textural. Order of features is performed using the principal component analysis (PCA). Then, the performance of three classifiers probabilistic neural network (PNN) and support vector machine (SVM) and Random Forest Tree is obtained. In total, the results show that the proposed method is accurate and sufficient to be applied in hematological laboratories. The average accuracy of segmentation was 98.98% and classification was 99.6%.

Classification of lymphocytes, monocytes, eosinophils, and neutrophils on white blood cells using hybrid Alexnet-GoogleNet-SVM

“…Classification was carried out with probabilistic neural network (PNN) and support vector machine and random Forest Tree. The accuracy obtained was 99.6% [24].…”

Identifikasi Penderita COVID-19 Berdasarkan Chest X-Ray Menggunakan Algoritma Jaringan Syaraf Tiruan Backpropagation

“…Beberapa peneliti telah melakukan penelitian citra medis khususnya paru menggunakan pendekatan algoritma jaringan syaraf tiruan Backpropagation. Penelitian yang dilakukan adalah deteksi kangker paru berdasarkan sampel jaringan atau biopsi paru yang di ekstraksi dengan pendekatan GLCM dan mendapatkan nilai akurasi data uji sebesar 81,25% [8]. Pendeteksian kangker paru berdasarkan citra Computed tomography (CT) dengan nilai akurasi 80% [9].…”

Extrinsic parameter's adjustment and potential implications in Plasmodium falciparum malaria diagnosis

“…Also, it has the potential to impact classification accuracy in computer‐assisted diagnosis. Although promising work has been focused on Machine learning (Abbas et al, 2016; Alqudah, Al‐Ta'ani, & Al‐Badarneh, 2018; Bashar, 2019; Devi et al, 2018; Fatima & Farid, 2020; Kunwar et al, 2018; Linder et al, 2014; Memeu, 2014; Sampathila et al, 2018; Savkare & Narote, 2015; Shruti et al, 2014; Tantisatirapong & Preedanan, 2020; Vijayalakshmi & Rajesh Kanna, 2019) and deep‐learning learning (Dinesh Jackson Samuel & Rajesh Kanna, 2019b; Dinesh Jackson Samuel & Rajesh Kanna, 2020), the effect of an adjustable brightness control knob, color temperature setting during image acquisition and its associated impacts on identifying positive and negative malaria cases have yet to be investigated. To overcome the aforementioned issues and minimize the influence of varying image brightness and color temperature, the proposed work focuses on preprocessing of features concerned rather than preprocessing of raw input images.…”