White Blood Cell Imaging Research Articles

Fine-grained interactive attention learning for semi-supervised white blood cell classification

White blood cell (WBC) is an essential part of the human immune system. To diagnose blood diseases, hematologists have to think about the WBC information. For instance, the number of each type of WBCs often implies the health condition of the human body. Thus, the classification of white blood cell images plays a significant role in the medical diagnosis process. However, manual WBC inspection is time-consuming and labor-intensive for experts, which means automated classification methods are needed for WBC recognition. Another problem is that the traditional automatic recognition system needs a large amount of annotated medical images for training, which is highly costly. In this respect, the semi-supervised learning framework has recently been widely used for medical diagnosis due to its specificity, which can explore relevant information from massive unlabeled data. In this study, a novel semi-supervised white blood cell classification method is proposed, named by Fine-grained Interactive Attention Learning (FIAL). It consists of a Semi-Supervised Teacher-Student (SSTS) module and a Fine-Grained Interactive Attention (FGIA) mechanism. In detail, SSTS employs limited labeled WBC images and generates predicted probability vectors for a large amount of unlabeled WBC samples, like a human. After top-k selection in predicted probabilities, the efficient data can be exploited from unlabeled WBC images for training. With a very small amount of annotated WBC images, FIAL achieves an average accuracy of 93.2% on BCCD dataset when giving 75 labeled images for each category, which sufficiently elaborates our excellent capability on semi-supervised white blood cell image classification task.

WBC-AMNet: Automatic classification of WBC images using deep feature fusion network based on focalized attention mechanism.

The recognition and classification of White Blood Cell (WBC) play a remarkable role in blood-related diseases (i.e., leukemia, infections) diagnosis. For the highly similar morphology of different WBC subtypes, it is too confused to classify the WBC effectively and accurately for visual observation of blood cell smears. This paper proposes a Deep Convolutional Neural Network (DCNN) with feature fusion strategies, named WBC-AMNet, for automatically classifying WBC subtypes based on focalized attention mechanism. To obtain more localized attention of CNN, the fusion features of the first and the last convolutional layer are extracted by focalized attention mechanism combining Squeeze-and-Excitation (SE) and Gather-Excite (GE) modules. The new method performs successfully in classifying monocytes, neutrophils, lymphocytes, and eosinophils on the complex background with an overall accuracy of 95.66%, better than that of general CNNs. The multi-classification accuracy of WBC-AMNet with the background segmentation is over 98% in all cases. In addition, Gradient-weighted Class Activation Mapping (Grad-CAM) is employed to visualize the attention heatmaps of different feature maps.

A large dataset of white blood cells containing cell locations and types, along with segmented nuclei and cytoplasm

Accurate and early detection of anomalies in peripheral white blood cells plays a crucial role in the evaluation of well-being in individuals and the diagnosis and prognosis of hematologic diseases. For example, some blood disorders and immune system-related diseases are diagnosed by the differential count of white blood cells, which is one of the common laboratory tests. Data is one of the most important ingredients in the development and testing of many commercial and successful automatic or semi-automatic systems. To this end, this study introduces a free access dataset of normal peripheral white blood cells called Raabin-WBC containing about 40,000 images of white blood cells and color spots. For ensuring the validity of the data, a significant number of cells were labeled by two experts. Also, the ground truths of the nuclei and cytoplasm are extracted for 1145 selected cells. To provide the necessary diversity, various smears have been imaged, and two different cameras and two different microscopes were used. We did some preliminary deep learning experiments on Raabin-WBC to demonstrate how the generalization power of machine learning methods, especially deep neural networks, can be affected by the mentioned diversity. Raabin-WBC as a public data in the field of health can be used for the model development and testing in different machine learning tasks including classification, detection, segmentation, and localization.

White Blood Cells Image Classification Based on Radiomics and Deep Learning

White blood cell (WBC) identification performance is highly correlated with the quality of the extracted features, with radiomic features having greater resolution and more detailed information and deep features having more robust semantic information. This research integrates these two aspects to creatively suggest a WBC classification model based on radiomics and deep learning. This research suggests a brand-new method for extracting radiomic features from WBC images as well as a dual-branch feature fusion network RCTNet based on CNN and Transformer for extracting deep features. The radiomic feature extraction method not only has a simple segmentation algorithm but also solves the problem of cell adhesion, can obtain higher quality shape, color, and texture features without segmenting intact cells, and is more generalizable. RCTNet can extract more critical and recognizable deep features from WBC nuclei, avoiding the influence of too much redundant and invalid information on the results, and has better performance than several existing CNN models. We compared the classification outcomes based on radiomics with those based on deep learning to confirm the efficacy of the WBC classification model based on radiomics and deep learning suggested in this research. The experimental results demonstrated that combining radiomic features and deep features significantly improved the classification accuracy, with an AUC exceeding 0.9995, accuracy, sensitivity, specificity, precision, and F1-score reaching 0.9880, 0.9823, 0.9883, 0.9968, and 0.9881, respectively. The model has significant research significance in clinical applications and aids physicians in improving diagnosis and screening for diseases of WBCs.

Exploiting the Multiscale Information Fusion Capabilities for Aiding the Leukemia Diagnosis Through White Blood Cells Segmentation

Leukemia is one of the most terminal types of blood cancer, and many people suffer from it every year. White blood cells (WBCs) have a significant association with leukemia diagnosis. Research studies reported that leukemia brings changes in WBC count and morphology. WBC accurate segmentation enables to detect morphology and WBC count which consequently helps in the diagnosis and prognosis of leukemia. Manual WBC assessment methods are tedious, subjective, and less accurate. To overcome these problems, we propose a multi-scale information fusion network (MIF-Net) for WBC segmentation. MIF-Net is a shallow architecture with internal and external spatial information fusion mechanisms. In WBC images, the cytoplasm is with low contrast compared to the background, whereas nuclei shape can be complex with an indistinctive boundary for some cases, therefore accurate segmentation becomes challenging. Spatial features in the initial layers of the network include fine boundary information and MIF-Net splits and propagates this boundary information on multi-scale for external information fusion. Multi-scale information fusion in our network helps in preserving boundary information and contributes to segmentation performance improvement. MIF-Net also uses internal information fusion after intervals for feature empowerment in different stages of the network. We evaluated our network for four publicly available datasets and achieved state-of-the-art segmentation performance. In addition, the proposed architecture exhibits superior computational efficiency by using only 2.67 million trainable parameters.

Enhancement of White Blood Cells Images using Shock Filtering Equation for Classification Problem

Medical image processing has developed rapidly in the last decade. The autodetection and classification of white blood cells (WBC) is one of the medical image processing applications. The analysis of WBC images has engaged researchers from medical also technology fields. Since WBC detection plays an essential role in the medical field, this paper presents a system for distinguishing and classifying WBC types: eosinophils, neutrophils, lymphocytes, and monocytes, using K-Nearest Neighbor (K-NN) and Logistic Regression (LR). This study aims to find the best accuracy of pre-processing images using original grayscale, shock filtering, and thresholding grayscale. The highest average accuracy in classifying WBC images in the conducting research is 43.54% using the LR algorithm from 2103 images. It is obtained from the combination of thresholding grayscale image and shock filtering equation to enhance the quality of an image. Overall, using two algorithms, KNN and LR, the classification accuracy can increase up to 12%.

WBC-based segmentation and classification on microscopic images: a minor improvement

Introduction White blood cells (WBCs) are immunity cells which fight against viruses and bacteria in the human body. Microscope images of captured WBCs for processing and analysis are important to interpret the body condition. At present, there is no robust automated method to segment and classify WBCs images with high accuracy. This paper aims to improve on WBCs image segmentation and classification method. Methods A triple thresholding method was proposed to segment the WBCs; meanwhile, a convolutional neural network (CNN)-based binary classification model that adopts transfer learning technique was proposed to detect and classify WBCs as a healthy or a malignant. The input dataset of this research work is the Acute Lymphoblastic Leukemia Image Database (ALL-IDB). The process first converts the captured microscope images into HSV format for obtaining the H component. Otsu thresholding is applied to segment the WBC area. A 13 × 13 kernel with two iterations was used to apply morphological opening on image to ameliorate output results. Collected cell masks were used to detect the contour of each cell on the original image. To classify WBCs into a healthy or a malignant category, characteristics and conditions of WBCs are to be examined. A transfer learning technique and pre-trained InceptionV3 model were employed to extract the features from the images for classification. Results The proposed WBCs segmentation method yields 90.45% accuracy, 83.81% of the structural similarity index, 76.25% of the dice similarity coefficient, and is computationally efficient. The accuracy of fine-tuned classifier model for training, validation and test sets are 93.27%, 92.31% and 96.15% respectively. The obtained results are high in accuracy and precision are over 96% and with lower loss value. Discussion Triple thresholding outperforms K-means clustering in segmenting smaller dataset. Pre-trained InceptionV3 model and transfer learning improve the flexibility and ability of classifier.

Multimodality Imaging in the Diagnosis of Prosthetic Valve Endocarditis: A Brief Review.

Infective endocarditis is a common and treatable condition that carries a high mortality rate. Currently the workup of infective endocarditis relies on the integration of clinical, microbiological and echocardiographic data through the use of the modified Duke criteria (MDC). However, in cases of prosthetic valve endocarditis (PVE) echocardiography can be normal or non-diagnostic in a high proportion of cases leading to decreased sensitivity for the MDC. Evolving multimodality imaging techniques including leukocyte scintigraphy (white blood cell imaging), 18F-fluorodeoxyglucose positron emission tomography (FDG-PET), multidetector computed tomographic angiography (MDCTA), and cardiac magnetic resonance imaging (CMRI) may each augment the standard workup of PVE and increase diagnostic accuracy. While further studies are necessary to clarify the ideal role for each of these imaging techniques, nevertheless, these modalities hold promise in determining the diagnosis, prognosis, and care of PVE. We start by presenting a clinical vignette, then evidence supporting various modality strategies, balanced by limitations, and review of formal guidelines, when available. The article ends with the authors' summary of future directions and case conclusion.

Improving Medical Image Decision-Making by Leveraging Metacognitive Processes and Representational Similarity.

Improving the accuracy of medical image interpretation can improve the diagnosis of numerous diseases. We compared different approaches to aggregating repeated decisions about medical images to improve the accuracy of a single decision maker. We tested our algorithms on data from both novices (undergraduates) and experts (medical professionals). Participants viewed images of white blood cells and made decisions about whether the cells were cancerous or not. Each image was shown twice to the participants and their corresponding confidence judgments were collected. The maximum confidence slating (MCS) algorithm leverages metacognitive abilities to consider the more confident response in the pair of responses as the more accurate "final response" (Koriat, 2012), and it has previously been shown to improve accuracy on our task for both novices and experts (Hasan etal., 2021). We compared MCS to similarity-based aggregation (SBA) algorithms where the responses made by the same participant on similar images are pooled together to generate the "final response." We determined similarity by using two different neural networks where one of the networks had been trained on white blood cells and the other had not. We show that SBA improves performance for novices even when the neural network had no specific training on white blood cell images. Using an informative representation (i.e., network trained on white blood cells) allowed one to aggregate over more neighbors and further boosted the performance of novices. However, SBA failed to improve the performance for experts even with the informative representation. This difference in efficacy of the SBA suggests different decision mechanisms for novices andexperts.

WBC-based segmentation and classification on microscopic images: a minor improvement.

Introduction White blood cells (WBCs) are immunity cells which fight against viruses and bacteria in the human body. Microscope images of captured WBCs for processing and analysis are important to interpret the body condition. At present, there is no robust automated method to segment and classify WBCs images with high accuracy. This paper aims to improve on WBCs image segmentation and classification method. Methods A triple thresholding method was proposed to segment the WBCs; meanwhile, a convolutional neural network (CNN)-based binary classification model that adopts transfer learning technique was proposed to detect and classify WBCs as a healthy or a malignant. The input dataset of this research work is the Acute Lymphoblastic Leukemia Image Database (ALL-IDB). The process first converts the captured microscope images into HSV format for obtaining the H component. Otsu thresholding is applied to segment the WBC area. A 13 × 13 kernel with two iterations was used to apply morphological opening on image to ameliorate output results. Collected cell masks were used to detect the contour of each cell on the original image. To classify WBCs into a healthy or a malignant category, characteristics and conditions of WBCs are to be examined. A transfer learning technique and pre-trained InceptionV3 model were employed to extract the features from the images for classification. Results The proposed WBCs segmentation method yields 90.45% accuracy, 83.81% of the structural similarity index, 76.25% of the dice similarity coefficient, and is computationally efficient. The accuracy of fine-tuned classifier model for training, validation and test sets are 93.27%, 92.31% and 96.15% respectively. The obtained results are high in accuracy and precision are over 96% and with lower loss value. Discussion Triple thresholding outperforms K-means clustering in segmenting smaller dataset. Pre-trained InceptionV3 model and transfer learning improve the flexibility and ability of classifier.

Classification of Blood Cells from Blood Cell Images Using Dense Convolutional Network

Classification of white blood cells plays a significant role in the detection of diseases which are infections caused by abnormalities in the immune system, allergies, anemia, leukemia, cancer, AIDS, etc. In traditional methods, experts manually examine the white blood cells under the microscope, and since this process is tedious, takes more time and can be more error-prone, automated systems have become necessary. Making this classification automatically will help experts in the detection of diseases. In this study, blood cells are classified from blood cell images using dense convolutional neural network. This paper intends to utilize a dense convolutional neural network model to overcome the blood cell classification problem that is one of the most compelling problems in blood diagnosis. In this study, a DenseNet121 model is built to classify blood cell images. Experiments are conducted on an open-access BCCD dataset (12507 white blood cell images that contain four types of white blood cells). Performance evaluations are performed on the accuracy of the techniques, and the results are compared with state-of-the-art deep learning-based approaches as Xception, VGG19, EfficientNetB1. In experimental studies, the highest accuracy (94%) is obtained with the proposed DenseNet121 model. Considering the high accuracy value obtained with this model, automatic detection of which class the cells belong to will speed up the diagnosis and allow more data to be examined by doctors.

A deep learning method for counting white blood cells in bone marrow images

BackgroundDifferentiating and counting various types of white blood cells (WBC) in bone marrow smears allows the detection of infection, anemia, and leukemia or analysis of a process of treatment. However, manually locating, identifying, and counting the different classes of WBC is time-consuming and fatiguing. Classification and counting accuracy depends on the capability and experience of operators.ResultsThis paper uses a deep learning method to count cells in color bone marrow microscopic images automatically. The proposed method uses a Faster RCNN and a Feature Pyramid Network to construct a system that deals with various illumination levels and accounts for color components' stability. The dataset of The Second Affiliated Hospital of Zhejiang University is used to train and test.ConclusionsThe experiments test the effectiveness of the proposed white blood cell classification system using a total of 609 white blood cell images with a resolution of 2560 × 1920. The highest overall correct recognition rate could reach 98.8% accuracy. The experimental results show that the proposed system is comparable to some state-of-art systems. A user interface allows pathologists to operate the system easily.

White Blood Cell Detection and Segmentation Combined with the Channel Space Weighted Feature Pyramid Networks

<p indent=0mm>The accurate detection and segmentation of leukocytes is a challenging task in medical image processing. The white blood cell image obtained under microscope is easily affected by impurities. There are many kinds of white cells, different shapes and small differences among them, and also exiting overlapping and adhesion phenomena, which leads to the inaccuracy of cell edge segmentation. The above problems are always the difficulties of detection and segmentation of white cell image. To solve the above problems, a leukocyte detection method based on Mask R-CNN and attention mechanism multi-scale feature fusion is proposed. Based on the Mask R-CNN structure, proposed thesis integrates the attention mechanism module in the FPN (feature pyramid networks) module, and proposes the CSFPN (channel spatial feature pyramid networks) structure. This structure can learn the weight of important channel features and the representation of important feature regions in the feature maps. At the same time, Skip-FPN module is added to the network structure, which fuses more low-level detailed information of leukocytes through short connection, so as to detect and segment leukocytes more accurately. Experimental results show that this method has good detection and segmentation performance. Under the Kaggle open source data set, the mAP value of this method for white blood cell detection reaches 98.25%, which has an increase of 1.25% compared with the previous improvement. The accuracy mIoU value reached 89.3%, which has an increase of 0.002% compared to the before improvement.

Pemanfaatan Scale Invariant Feature Transform Berbasis Saliency untuk Klasifikasi Sel Darah Putih

White blood cells are cells that makeup blood components that function to fight various diseases from the body (immune system). White blood cells are divided into five types, namely basophils, eosinophils, neutrophils, lymphocytes, and monocytes. Detection of white blood cell types is done in a laboratory which requires more effort and time. One solution that can be done is to use machine learning such as Support Vector Machine (SVM) with Scale Invariant Feature Transform (SIFT) feature extraction. This study uses a dataset of white blood cell images that previously carried out a pre-processing stage consisting of cropping, resizing, and saliency. The saliency method can take a significant part in image data and. The SIFT feature extraction method can provide the location of the keypoint points that SVM can use in studying and recognizing white blood cell objects. The use of region-contrast saliency with kernel radial basis function (RBF) yields the best accuracy, precision, and recall results. Based on the test results obtained in this study, saliency can improve the accuracy, precision, and recall of SVM on the white blood cell image dataset compared to without saliency.

Perforated Appendicitis During a Pandemic: The Downstream Effect of COVID-19 in Children

IntroductionCoronavirus Disease-19 (COVID-19) was declared a pandemic in March 2020. States issued stay-at-home orders and hospitals cancelled non-emergent surgeries. During this time, we anecdotally noticed more admissions for perforated appendicitis. Therefore, we hypothesized that during the months following the COVID-19 pandemic declaration, more children were presenting with perforated appendicitis.Materials and MethodsThis is a retrospective cohort study reviewing pediatric patients admitted at a single institution with acute and/or perforated appendicitis between October 2019 to May 2020. Interval appendectomies were excluded. COVID-19 months were designated as March, April, and May 2020. Additional analysis of March, April, and May 2019 was performed for comparison purposes. Analyzed data included demographics, symptoms, white blood cell count, imaging findings, procedures performed, and perforation status. Statistical analysis was performed.ResultsDuring the study period, 285 patients were admitted with the diagnosis of acute appendicitis with 95 patients being perforated. We identified a significant increase in perforated appendicitis cases in the three COVID-19 months compared with the preceding five months (45.6% vs 26.4%; P <0.001). In addition, a similar significant increase was identified when comparing to the same months a year prior (P = 0.003). No significant difference in duration of pain was identified (P=0.926).ConclusionThe COVID-19 pandemic and its associated stay-at-home orders have had downstream effects on healthcare. Our review has demonstrated a significant increase in the number of children presenting with perforated appendicitis following these stay-at-home ordinances. These results demonstrate that further investigations into the issues surrounding access to healthcare, especially during this pandemic, are warranted.

Hybrid Inception v3 XGBoost Model for Acute Lymphoblastic Leukemia Classification

Acute lymphoblastic leukemia (ALL) is the most common type of pediatric malignancy which accounts for 25% of all pediatric cancers. It is a life-threatening disease which if left untreated can cause death within a few weeks. Many computerized methods have been proposed for the detection of ALL from microscopic cell images. In this paper, we propose a hybrid Inception v3 XGBoost model for the classification of acute lymphoblastic leukemia (ALL) from microscopic white blood cell images. In the proposed model, Inception v3 acts as the image feature extractor and the XGBoost model acts as the classification head. Experiments indicate that the proposed model performs better than the other methods identified in literature. The proposed hybrid model achieves a weighted F1 score of 0.986. Through experiments, we demonstrate that using an XGBoost classification head instead of a softmax classification head improves classification performance for this dataset for several different CNN backbones (feature extractors). We also visualize the attention map of the features extracted by Inception v3 to interpret the features learnt by the proposed model.

Identification of White Blood Cells Using Machine Learning Classification Based on Feature Extraction

In various disease diagnoses, one of the parameters is white blood cells, consisting of eosinophils, basophils, neutrophils, lymphocytes, and monocytes. Manual identification takes a long time and tends to be subjective depending on the staff's experience, so the automatic identification of white blood cells will be faster and more accurate. White blood cells are identified by examining a colored blood smear (SADT) and examined under a digital microscope to obtain a cell image. Image identification of white blood cells is determined through HSV color space segmentation (Hue, Saturation Value) and feature extraction of the Gray Level Cooccurrence Matrix (GLCM) method using the Angular Second Moment (ASM), Contrast, Entropy, and Inverse Different Moment (IDM) features. The purpose of this study was to identify white blood cells by comparing the classification accuracy of the K-nearest neighbor (KNN), Naïve Bayes Classification (NBC), and Multilayer Perceptron (MLP) methods. The classification results of 100 training data and 50 white blood cell image testing data. Tests on the KNN, NBC, and MLP methods yielded an accuracy of 82%, 80%, and 94%, respectively. Therefore, MLP was chosen as the best classification model in the identification of white blood cells.

448-P: Serum Inflammatory Markers Are Not Correlated with Detection of Osteomyelitis by 99mTC-WBC SPECT/CT Imaging in Subjects with Diabetes and Foot Infection

Osteomyelitis is a common complication in diabetic foot ulcers. Accurate and early detection of diabetic foot osteomyelitis (DFO) is key to optimizing clinical outcomes and avoiding amputation. Using bone biopsy histopathology as the gold diagnostic standard, we evaluate the sensitivity and specificity of radiolabeled white blood cell imaging combined with CT (99mTc-WBC SPECT/CT) for the evaluation of DFO. We also assess the ability of the inflammatory markers CRP and ESR to predict osteomyelitis diagnosis by 99mTc-WBC SPECT/CT imaging. Subjects with diabetes and foot ulcerations presenting to a large tertiary care facility were examined by podiatry. Of the 48 subjects who consented to participation, 44 completed imaging, labwork, and underwent biopsy. Serum CRP and ESR were collected within three days of consent. In all cases, WBC imaging studies were performed prior to biopsy. The WBC SPECT/CT scans were read by an experienced Nuclear Medicine radiologist. Pathology was performed in the hospital laboratory and interpreted by a pathologist experienced in MSK infections. SPECT/CT and pathology reads were coded as either positive or negative for osteomyelitis. Sensitivity and specificity of 99mTc-WBC SPECT/CT for DFO diagnosis was 90.9% and 73.7%, respectively. WBC imaging had a PPV of 80.0%, and 87.5% NPV (n=41). CRP and ESR serum levels were grouped according to positive or negative DFO diagnosis as determined by WBC scan. Neither CRP (p=0.9895) nor ESR levels (p=0.8382) were correlated with DFO diagnosis based on WBC imaging results. Using bone biopsy results as the reference standard, radiolabeled 99mTc-WBC SPECT/CT demonstrates sensitivity for DFO. However, CRP and ESR were not found to be predictive in the diagnostic interpretation of 99mTc-WBC SPECT/CT imaging for the diagnosis of DFO. Disclosure A. Sherwood: None. L. A. Lavery: None. O. K. Oz: n/a. K. L. Rubitschung: None. A. L. Killeen: None. P. Crisologo: None. K. Grigoropoulos: None. A. Pajouh: None. J. Kep: None. D. Wukich: Consultant; Self; Orthofix Medical Inc., Wright Medical Group, Other Relationship; Self; Arthrex, Inc. J. La fontaine: None. Funding American Diabetes Association (1-17-ICTS-056 to O.K.O.)

Attention-Aware Residual Network Based Manifold Learning for White Blood Cells Classification.

The classification of six types of white blood cells (WBCs) is considered essential for leukemia diagnosis, while the classification is labor-intensive and strict with the clinical experience. To relieve the complicated process with an efficient and automatic method, we propose the Attention-aware Residual Network based Manifold Learning model (ARML) to classify WBCs. The proposed ARML model leverages the adaptive attention-aware residual learning to exploit the category-relevant image-level features and strengthen the first-order feature representation ability. To learn more discriminatory information than the first-order ones, the second-order features are characterized. Afterwards, ARML encodes both the first- and second-order features with Gaussian embedding into the Riemannian manifold to learn the underlying non-linear structure of the features for classification. ARML can be trained in an end-to-end fashion, and the learnable parameters are iteratively optimized. 10800 WBCs images (1800 images for each type) is collected, 9000 images and five-fold cross-validation are used for training and validation of the model, while additional 1800 images for testing. The results show that ARML achieving average classification accuracy of 0.953 outperforms other state-of-the-art methods with fewer trainable parameters. In the ablation study, ARML achieves improved accuracy against its three variants: without manifold learning (AR), without attention-aware learning (RML), and AR without attention-aware learning. The t-SNE results illustrate that ARML has learned more distinguishable features than the comparison methods, which benefits the WBCs classification. ARML provides a clinically feasible WBCs classification solution for leukemia diagnose with an efficient manner.

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

White blood cells (WBC), which form the basis of the immune system, protect the body from foreign invaders and infectious diseases. While the number and structural features of WBCs can provide important information about the health of people, the ratio of the subtypes of these cells and observable deformations are a good indicator in the diagnostic process. The recognition of cells of the type of lymphocytes, neutrophils, eosinophils, basophils and monocytes is critical. In this article, Deep Learning based Hybrid CNN (Convololutional Neural Network) model is proposed for classification of eosinophils, lymphocytes, monocytes, and neutrophils WBCs. The model presented is based on pretrained Alexnet and Googlenet architectures. The feature vector in the last pooling layer of both CNN architectures has been merged, and the resulting feature vector is classified by the Support Vector Machine. To determine the superiority of the proposed method, the classification was also performed and compared using pretrained Alexnet and Googlenet. Hybrid Alexnet-Googlenet-SVM model provides higher accuracy than pretrained Alexnet and Googlenet. The proposed method has been tested with WBC images from Kaggle and LISC database. Accuracy and F1-score were 99.73%, 0.99 and 98.23%, 0.98 for both data sets, respectively.