Leucocyte classification for leukaemia detection using image processing techniques.

Automatic whole blood cell analysis from blood smear using label-free multi-modal imaging with deep neural networks

3D microscopy imaging is important for understanding complex biological assemblies. Computers and digital image acquisition systems have made possible the 3D visualization of images obtained from optical microscopes. This paper discussed a morphological approach for 3D blood cell visualization. The 3D plotted image is obtained from 2D image by combining 2D image space with the intensity value presented as third axis. 2D image detection techniques are commonly apply for blood cell analysis on image taken from microscopy or scanner sample which normally could be affected by noise. In order to enhance the detection of blood cell for 2D image and to obtain the clearest view of the blood cell, 3D image detection, which is more flexible and accurate, is proposed. This approach helps to obtain a clear image for disease analysis. A color image taken from microscopy or scanner has to gone through such image analysis using intensity, and edge operator to detect the blood cell disease, before 3D visualization can be done. INTRODUCTION In blood analysis component of the blood are distinguished by their dimensions and color. Blood cell classification and analysis is an important diagnostic technique in the detection of many illnesses [1]. The importance of traceability in the medical diagnostics is a technique that can be determined from a microscope slide or from a set of images. An accurate diagnosis could aid in diagnoses treatment disease, therefore the accurate details of the cells has to be analytical. In malarial blood, the red corpuscles of vertebrates are infected by malarial parasites [2]. The morphological blood cell image analysis includes calculation of geometrical features for recognition of pathology [3]. The goal of this paper is to enhance the detection by using or viewing blood cells visualization to provide a clear view to pathologies in order to reduce noise of the image for more accuracy diagnoses. BLOOD IMAGE DETECTION TECHNIQUES There are automated technologies for cell analysis, such as impendence analysis and flow cytometry, these method lack the ability to visualize each of the cell used in there determinations, and visually inspect them to ensure that they were classified appropriately [4]. In edge image detection acquired from stained blood, smears are suitable for measuring and characterizing properties of individual cells [5]. Edge detection is essentially the operation of detection significantly local changes in image. In one dimension, a step edge is associated with local peak in the first derivative. The gradient is a measure of change in function, and an image can be considered to be an array of sample of some continues function of image intensity. The gradient is the two-dimensional equivalent of the first derivative and is defined as the vector. 3D Visualization For Blood Cells Analysis Versus Edge Detection

Large number of diseases can be diagnosed by the counting and classification of blood cells. One type of the most common blood diseases is Acute Lymphoblastic Leukemia (ALL). Rapid and uncontrolled growth of immature leukemic cells (also named as blast cells) is used to characterize ALL. The morphological analysis of the bone marrow and blood smear is the primary step of diagnosis. Valuable information is provided by the features extracted from the blood cells for the confirmation of the diagnosis. As the process have numerous drawbacks, a computerized system will be helpful for the analysis of stained microscopic images of the blood cells for quantitative examination. This paper demonstrates a new and simple segmentation approach using watershed transform and two simpler classification models to diagnose the presence of ALL in blood smear accurately.

Among the blood cells, white blood cells (WBC), which play a crucial role in forming our body's defense system, are essential components. Originating in the bone marrow, these cells serve as the fundamental components of the immune system, shouldering the responsibility of safeguarding the body against foreign microbes and diseases. Insufficient WBC counts may compromise the body's skill to resist infections, a status known as leukopenia. White blood cell counting is a specialty procedure that is usually carried out by qualified physicians and radiologists. Thanks to recent advances, image processing techniques are frequently used in biological systems to identify a wide spectrum of illnesses. In this work, image processing techniques were applied to enhance the white blood cell deep learning models' classification accuracy. To expedite the classification process, Convolutional Neural Network (CNN) models were combined with Ridge feature selection and Maximal Information Coefficient techniques. These tactics successfully determined the most important characteristics. The selected feature set was then applied to the classification procedure. ResNet-50, VGG19, and our suggested model were used as feature extractors in this study. The categorizing of white blood cells was completed with an amazing 98.27% success rate. Results from the experiments demonstrated a considerable improvement in classification accuracy using the proposed CNN model.

Counting and analysis of blood cells allow the evaluation and diagnosis of a vast number of diseases. In particular, the identification of white blood cells and red blood cells is a topic of great interest to hematologists. Nowadays the observation of blood samples is still performed manually by skilled operators. This task is tedious, lengthy and repetitive, and the results accuracy heavily depends on the operator skills. Differently, the automated analysis by computer is performed quickly, requires only one image of the blood sample and provides precise results. One of the major steps on image analysis is segmentation, that subdivides images into meaningful parts. Thresholding is one of the most used technique and subdivides the image pixels on the basis of their intensity
\ngrey levels. There are many crisp techniques for calculating the threshold value of an image. Recently many intuitionistic fuzzy methods have been proposed to determine the optimal threshold value, showing better results for segmentation but not even better computational performance. In this paper we propose an intuitionistic fuzzy set approach for optimal threshold selection based on computations performed on the histogram. This method is then extended in order to perform multiple thresholds and in order to take into account possible local variations on the image. The proposed approach has been tested on peripheral blood images, in order to subdivide the various image components, showing excellent performance both for segmentation both in terms of speed.

Background: Leukemia is a malignant disease of the blood and bone marrow that leads to cancer and accounts for 2.5% of all other cancers, with a small mortality rate (3.1) among the deaths of other cancers. Objective: This paper discusses the global data on leukemia-related cancer, its analysis and comparison by region, and a discussion of children’s cancer cases of this type, in addition to an Iraqi leukemia patient analysis of white blood cell (WBC) count and hemoglobin (HB) levels before and after cancer collection from laboratories in medical city hospitals. Materials & Methods: Publications collected from the WHO for the year 2022 on the prevalence and mortality of leukemia-related diseases by region were used, with comparisons made and their indicators identified, as illustrated with graphs for each country. The blood parameters (WBC and Hb) of 25 leukemia patients before and after cancer were analyzed via paired sample t tests. Results: The data analysis shows that this type of cancer has the lowest incidence and mortality compared to other cancer types; Oceania has the lowest incidence and mortality, whereas Asia and Europe have the highest. Rates and deaths among children younger than nine years. In childhood deaths, there are two types of leukemia: acute lymphoblastic leukemia, precursor cell lymphoblastic leukemia, and NOS leukemia. The two types are equally likely. Two types, acute and precursor cell lymphoblastic leukemia, were present in a lower percentage of girls than boys, whereas other types were present in higher percentages. prevalence. In 2022, the number of cases of leukemia increased, particularly among Asian men. Conclusion: This paper offers a comprehensive explanation of leukemia by analyzing mortality and prevalence data and providing information on the frequency of this disease in children under ten years of age.

As noted in a review paper,1 a large-scale study revealed that approximately 1.1%–3.8% of all children with Down syndrome (DS) are born with mosaic DS.2, 3 DS is diagnosed postnatally by chromosome G-banding analysis of peripheral blood cultures, and most patients do not undergo subsequent repeat analysis. We present secular changes in the mosaic ratio of six patients with mosaic DS. The G-banding, chromosome 21 fluorescence in situ hybridization (chr21 FISH) from peripheral blood cultures and the chr21 FISH analysis of uncultured blood cells showed a drastic reduction in the mosaic ratio in five patients aged between 7 and 23 years. The mosaic ratio of a child aged 3 years and 3 months gradually, but not drastically, decreased over time. However, no decrease in the mosaic ratio was seen in the buccal mucosa or fibroblasts by chr21 FISH in any mosaic patient. Low-invasive chromosomal analyses of 100 cells per patient from different tissues were performed by SRL, Inc. The analyses were as follows (Table 1). (1) T lymphocyte chromosome analyses of cultured peripheral blood samples by PHA (phytohemagglutinin) stimulation, including (1-a) G-banding and (1-b) chr21 FISH analysis using chromosome 21q22.13-q22.2 probes (Vysis, Abbott Molecular); (2) chr21 FISH analysis of uncultured blood cells, including various leukocytes, instead of bone marrow puncture; (3) chr21 FISH analysis of epithelial cells from the buccal mucosa; and (4) chr21 FISH analysis of fibroblasts cultured from a tooth extraction. The six patients (A–F) with mosaic DS (five females and one male) were aged from 3 years and 3 months to 23 years (Table 1). Five patients (B–F) aged between 7 and 23 years showed drastically decreased mosaic ratios (1%–8%) by G-banding, and FISH analysis of the peripheral blood culture and of uncultured blood cells compared with the postnatal result (Table 1, the red font). Patient A showed a gradual decrease in mosaicism on G-banding and FISH of the peripheral blood culture and by FISH analysis of uncultured blood cells at 11 months, 2 years and 9 months, and 3 years and 3 months of age compared with the postnatal result (Table 1, the red font). However, none of the six patients showed a decreased mosaic ratio in the buccal mucosa analysis or tooth-extraction-derived-fibroblast FISH analysis. Previously, two reports documented a decreased mosaic ratio in mosaic DS patients during early childhood, mainly in peripheral blood cultures.4, 5 We studied longitudinal changes in the mosaic ratio of six mosaic DS patients, including young adults. No decrease was observed in the epithelial cells of the buccal mucosa or cultured fibroblasts. The observed secular decrease in the mosaic ratio in cultured and uncultured peripheral blood cells did not appear to be a mechanism of trisomy rescue because six adult patients with standard trisomy 21 and one patient with translocated trisomy 21 showed no decrease in trisomy cell number in cultured and uncultured peripheral blood cells over time (Table S1, the green font). Within the bone marrow, hematopoietic stem cells exist in a hypoxic microenvironment (niche) and are in a quiescent (G0), undifferentiated state.6 In this microenvironment, trisomy cells are considered more detrimental to viability than normal cells, which may explain the secular decrease in chromosomal mosaicism with mosaic DS in this study. Permission has been obtained from the patients' mothers for the participation in this study. Financial support was provided by JSPS KAKENHI (Grant number: JP18K02497). The authors declare no conflict of interest. This study has received ethical approval from Tokyo Kasei University Ethics Committee (Ita h-29-5) and the Tokyo Metropolitan Tobu Medical Center for Children with Developmental Disabilities Ethics Committee (29MoriTobuCe256-1). TABLE S1. Chromosomal analyses of the trisomy ratio of eight patients, seven with standard trisomy 21 and one translocated trisomy 21 (patient H). Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Blood cell classification and counting are vital for the diagnosis of various blood-related diseases, such as anemia, leukemia, lymphoma, and thrombocytopenia. The manual process of blood cell classification and counting is time-consuming, prone to errors, and labor-intensive. Therefore, we have proposed a deep learning (DL)-based automated system for blood cell classification and counting from microscopic blood smear images. We classify a total of nine types of blood cells, including Erythrocyte, Erythroblast, Neutrophil, Basophil, Eosinophil, Lymphocyte, Monocyte, Immature Granulocytes, and Platelet. Several preprocessing steps like image resizing, rescaling, contrast enhancement and augmentation are utilized. To segment the blood cells from the entire microscopic images, we employed the U-Net model. This segmentation technique aids in extracting the region of interest (ROI) by removing complex and noisy background elements. Both pixel-level metrics such as accuracy, precision, and sensitivity, and object-level evaluation metrics like Intersection over Union (IOU) and Dice coefficient are considered to comprehensively evaluate the performance of the U-Net model. The segmentation model achieved impressive performance metrics, including 98.23% accuracy, 98.40% precision, 98.26% sensitivity, 95.97% Intersection over Union (IOU), and 97.92% Dice coefficient. Subsequently, a watershed algorithm is applied to the segmented images to separate overlapped blood cells and extract individual cells. We have proposed a BloodCell-Net approach incorporated with custom light weight convolutional neural network (LWCNN) for classifying individual blood cells into nine types. Comprehensive evaluation of the classifier’s performance is conducted using metrics including accuracy, precision, recall, and F1 score. The classifier achieved an average accuracy of 97.10%, precision of 97.19%, recall of 97.01%, and F1 score of 97.10%. A 5-fold cross-validation technique is applied to split the data, which not only aids in reducing overfitting but also helps in generalizing the model.

The problem of identifying and counting blood cells within the blood smear is of both theoretical and practical interest. The differential counting of blood cells provides invaluable information to pathologist for diagnosis and treatment of many diseases. In this paper we propose an efficient hierarchical blood cell image identification and classification method based on multi-class support vector machine. In this automated process, segmentation and classification of blood cells are the most important stages. We segment the stained blood cells in digital microscopic images and extract the geometric features for each segment to identify and classify the different types of blood cells. The experimental results are compared with the manual results obtained by the pathologist, and demonstrate the effectiveness of the proposed method.

Counting and analyzing of blood cells, as well as their subcellular structures, are indispensable for understanding biological processes, studying cell functions, and diagnosing diseases. In this paper, we combine digital holographic microscopy with cell segmentation guided by the Sobel operator using Dice coefficients for automatic threshold selection and aimed to automatic counting and analysis of blood cells in flow and different kinds of cells in the static state. We demonstrate the proposed method with automatic counting and analyzing rat red blood cells (RBCS) flowing in a microfluidic device, extracting quickly and accurately the size, concentration, and dry mass of the sample in a label-free manner. The proposed technique was also demonstrated for automatic segmentation of different cell types, such as COS7 and Siha. This method can help us in blood inspection, providing pathological information in disease diagnosis and treatment.

Artificial intelligence (AI) has emerged as a cutting-edge tool, simultaneously accelerating, securing, and enhancing the diagnosis and treatment of patients. An exemplification of this capability is evident in the analysis of peripheral blood smears (PBS). In university medical centers, hematologists routinely examine hundreds of PBS slides daily to validate or correct outcomes produced by advanced hematology analyzers assessing samples from potentially problematic patients. This process may logically lead to erroneous PBC readings, posing risks to patient health. AI functions as a transformative tool, significantly improving the accuracy and precision of readings and diagnoses. This study reshapes the parameters of blood cell classification, harnessing the capabilities of AI and broadening the scope from 5 to 11 specific blood cell categories with the challenging 11-class PBC dataset. This transformation facilitates a more profound exploration of blood cell diversity, surpassing prior constraints in medical image analysis. Our approach combines state-of-the-art deep learning techniques, including pre-trained ConvNets, ViTb16 models, and custom CNN architectures. We employ transfer learning, fine-tuning, and ensemble strategies, such as CBAM and Averaging ensembles, to achieve unprecedented accuracy and interpretability. Our fully fine-tuned EfficientNetV2 B0 model sets a new standard, with a macro-average precision, recall, and F1-score of 91%, 90%, and 90%, respectively, and an average accuracy of 93%. This breakthrough underscores the transformative potential of 11-class blood cell classification for more precise medical diagnoses. Moreover, our groundbreaking “Naturalize” augmentation technique produces remarkable results. The 2K-PBC dataset generated with “Naturalize” boasts a macro-average precision, recall, and F1-score of 97%, along with an average accuracy of 96% when leveraging the fully fine-tuned EfficientNetV2 B0 model. This innovation not only elevates classification performance but also addresses data scarcity and bias in medical deep learning. Our research marks a paradigm shift in blood cell classification, enabling more nuanced and insightful medical analyses. The “Naturalize” technique’s impact extends beyond blood cell classification, emphasizing the vital role of diverse and comprehensive datasets in advancing healthcare applications through deep learning.

Blood cell classification using interpretable shape features: A Comparative study of SVM models and CNN-Based approaches

Background of study: Blood cell analysis is vital for diagnosing medical conditions, but traditional manual methods are laborious and error-prone. Deep learning, especially YOLO models, offers automated solutions for medical image analysis. However, the real-world effectiveness of the latest YOLOv11 in blood cell detection is not thoroughly investigated, as general object detection improvements may not translate to biomedical images due to their unique characteristics.Aims and scope of paper: This study systematically compares YOLOv10 and YOLOv11 on a public blood cell detection dataset to assess if YOLOv11's advancements provide tangible benefits for blood cell classification. The goal is to identify the most effective model for accurate and efficient detection in microscopic images, guiding AI-driven diagnostic tool selection.Methods: Both models were trained and tested under identical conditions using the Kaggle Blood Cell Detection Dataset (RBCs, WBCs, Platelets). Images were resized to 640x640 pixels. Performance metrics included mAP (mAP@50 and mAP@50–95), Precision, Recall, F1-score, speed, model complexity, and training time.Result: YOLOv11n consistently showed higher accuracy (mAP50: 0.9279 vs. 0.9120; mAP50-95: 0.6524 vs. 0.6347), particularly for RBCs and WBCs. However, YOLOv11n had longer inference (11.35 ms/image) and postprocessing times (8.64 ms/image) compared to YOLOv10n (7.00 ms/image and 0.90 ms/image). YOLOv11n trained faster (0.311 hours vs. 0.375 hours), with a smaller model size (5.5 MB vs. 5.8 MB), fewer parameters, and reduced computational complexity.Conclusion: YOLOv11n offers superior accuracy and improved training efficiency, making it suitable for medical image object detection where precision is paramount. The increased inference and postprocessing times indicate a performance-speed trade-off. Model selection should balance these factors based on deployment context.

Improving blood cells classification in peripheral blood smears using enhanced incremental training

Background/Objectives: Accurate detection and classification of blood cell types in microscopic images are crucial for diagnosing various hematological conditions. This study aims to develop and evaluate advanced architectures for automating blood cell detection and classification using the newly proposed YOLOv10 and YOLOv11 models, with a specific focus on identifying red blood cells (RBCs), white blood cells (WBCs), and platelets in microscopic images as a preliminary step of the complete blood count (CBC). Methods: The Blood Cell Count Detection (BCCD) dataset was enriched using data augmentation techniques to improve model robustness and diversity. Extensive experiments were performed, including complete weight initialization, advanced optimization strategies, and meticulous hyperparameter tuning for the YOLOv11 architecture. Results: The YOLOv11-l model achieved an overall mean Average Precision (mAP) of 93.8%, reflecting its robust accuracy across multiple blood cell types. Conclusions: The findings underscore the efficacy of the YOLOv11 architecture in automating blood cell classification with high precision, demonstrating its potential to enhance hematological analyses and support clinical diagnosis.