Automated Anaemia Detection:Machine Learning

Approach

Ashwin M A K Hithesh Sathvi Rai

Department of Electronics & Department of Electronics & Department of Electronics &
Communication Engineering Communication Engineering Communication Engineering
Sahyadri College of Engineering & Sahyadri College of Engineering & Sahyadri College of Engineering &
Management,Adyar Management,Adyar Management,Adyar
Mangaluru,India Mangaluru,India Mangaluru,India

Chaithanya J Roopashree
Department of Electronics & Department of Electronics &
Communication Engineering Communication Engineering
Sahyadri College of Engineering & Sahyadri College of Engineering &
Management,Adyar Management,Adyar
Mangaluru,India Mangaluru,India

*Corresponding Author: Roopashree

Email Id:[email protected]

Abstract—Anemia, which is defined as a reduction in the careful inspection of blood smears. identifying anemia,
quantity of red blood cells or hemoglobin content, is a common which is a blood condition that is characterized by iron
worldwide health concern impacting millions of people. The deficiency in hemoglobin and red blood cells. In the medical
timely identification and precise diagnosis of anemia are industry, focus is crucial. All phases of life are affected by
essential for efficient treatment and averting related anemia, although pregnant women and young children are
consequences. The project entails creating an automated image more at risk. Anemia is a worldwide health issue that impacts
processing method to evaluate digital pictures of patient- 1.62 billion individuals, or roughly 24.8% of the global
provided blood smears. Red blood cell morphological
populace, ranking it as the second most reason for the
anomalies, such as differences in size, shape, and color, that are
sickness globally, as per the World Health Organization
suggestive of various forms of anemia, will be recognized and
quantified by the algorithm. Utilizing machine learning
(WHO). In order to meet the urgent requirement to identify
methodologies for both feature extraction and anemia, the analysis of pictures from peripheral blood smears
classification and train the system through an extensive (PBS) has long been used as a vital diagnostic instrument.
collection of annotated images of blood smears. To improve the Even while analysis is useful, it is tedious, prone to errors,
algorithm's diagnostic precision and prognostic powers, takes a lot of time, and needs the knowledge of qualified
clinical data such as patient demographics, medical histories, labs. The automation of PBS analysis has generated a
and laboratory test results will also be incorporated. The goal thriving field of study and motivated many research teams to
of the suggested method is to give medical professionals a look into cutting-edge methods that make use of image
dependable, affordable, and non-invasive tool for the early processing tools. This study starts with a thorough review. of
identification and categorization of anemia. This initiative has the techniques employed to examine the features of RBCs
the potential to greatly improve patient outcomes and lessen from PBS pictures, primarily emphasizing their utilization in
the burden of anemia-related consequences on healthcare identification of anemia.
systems around the world by utilizing developments in digital
imaging and machine learning technologies. The certain risk factors for blood cancers:
• Chronic infections: By interfering with the body's ability to
Keywords— Anemia, Red Blood Cell (RBC), Blood samples,
produce red-blood cells, long-term infections or illnesses such as
Neural Networks, Peripheral Blood Smear (PBS)
HIV/AIDS can lead to anemia.
I. INTRODUCTION
• Gender and age: Because of their fast growth and
Anemia is a chronic worldwide health problem that development, infants and teenagers may be more vulnerable.
affects individuals of all ages and is defined by a reduction in Age-appropriate women are more vulnerable because of
the quantity or quality of red blood cells. It is crucial to monthly blood loss.
identify anemia as soon as possible and accurately.
Nevertheless, existing diagnostic approaches often fail to • Nutritional deficiencies: Inadequate consumption of iron in
meet the accuracy necessary for subtle diagnosis and the diet, particularly for vegetarians or people with bad
efficient intervention. Our project seeks to address this eating habits.
urgent need to revolutionize the detection of anemia using • Family history: Genetic factors may be involved in
the cutting-edge method of "Detection of Anemia: Using illnesses such as thalassemia or sickle cell anemia.
Blood Smear Analysis." This research attempts to identify
minute morphological cues inside red blood cells through
 II. RELATED WORKS morphology of blood cells. Image processing techniques are
A. Image Processing-Based Approach used to augment and preprocess these images, enhancing
characteristics related to anemia for simpler identification.
There have been numerous other successful advances in The project also entails feature extraction, which includes
image processing in the healthcare industry and other measurements like cell size and red blood cell count. With
domains. Image processing is the process of carrying out machine learning techniques for automation, these features
standard procedures on an image to produce an improved or can be utilized for diagnosis and classification. To identify
high-quality picture. It's a technique for turning an image red blood cells (RBCs) and white blood cells (WBCs) in
into a digital shape. This technique has a wide range of microscopic pictures, the research [17] employs image
applications. sectors. There are three simple steps to it: processing techniques. The main objective is to develop a
utilizing an optical image analysis, a scanner for importing system that can recognize and differentiate these vital blood
images, and management, which incorporates methods for cell types for use in research and medical diagnostics. It is
compression, image enhancement, as well as finding highly likely that the study will entail gathering and
patterns in pictures like as photos from satellites. Image examining microscopic pictures, which will offer
processing is occasionally thought to be improper image comprehensive visual data regarding blood cell morphology.
editing in order to attain a specific standard of attractiveness To improve image quality, reduce noise, and get ready for
or to back a well-liked truth. By looking for abnormalities in cell detection, image processing techniques are applied.
red blood cell pictures, image processing is used to diagnose Segmentation algorithms are used to isolate individual cells
anemia. Common segmentation methods include clustering from the background, while feature extraction techniques
and marker controlled watershed approaches. To categorize are used to retrieve pertinent attributes including size, shape,
different kinds of anemia and subtypes; in particular, and colour. The study [19] focuses on the identification of
machine learning is employed Support Vector Machines aberrant red blood cells (RBCs) and the diagnosis of
(SVM). Automated segmentation of images Precision is particular forms of anemia using image processing methods
aided by methods like Otsu's approach. This An integrated and machine learning, most especially Support Vector
strategy increases efficiency and accuracy of the diagnosis Machines (SVM). The main objective of the project is to
of anemia, enabling patients to get more individualized create a system that can automatically identify anemia by
attention. The research [2] proposed approaches for differentiating between normal and aberrant red blood cells
classifying sickle cell anemia were compared: KNN, SVM, in microscopic blood samples. This process probably
and ELM. To identify red blood cells with a sickle cell involves collecting and analysing photographs of blood
formation illness patient, the researchers employed methods cells, which are subsequently improved and pre-processed
including geometric characteristics, Sobel edge detection, utilizing image processing methods to get them ready for in-
random walk, Fuzzy Cmeans clustering, and morphological depth cell analysis. The study [13] looks on the analysis of
filters. The approaches' reported average accuracy varied white blood cell (WBC) counting in blood smear pictures
from. From 85% to 95%. Red blood cells are impacted by using various color segmentation techniques. Researchers
sickle cell anemia. The blood cells are responsible for from University Tun Hussein Onn Malaysia are
transporting oxygen to the tissues within the body. It's investigating different methods to precisely detect and tally
imperative to identify this illness early. Thus, it may shorten white blood cells in digital blood smear pictures. The
one's life. In order to accomplish this, an image processing Department of Mechatronic and Robotic Engineering,
method is being created to determine the sickle cell presence Faculty of Electrical and Electronic Engineering, Research
in blood samples, including procedures like noise reduction, Centre for Applied Electromagnetics (EMcenter) is the
grayscale image conversion, and noise filtering. The goal of location of the study's operation. The goal is to improve the
the study [4] is to precisely separate high-resolution whole- precision and efficacy of WBC counting, which is essential
slide blood cells according to their semantic content using for the diagnosis of a variety of medical disorders, by testing
image processing techniques. The primary goal of this work various color segmentation techniques. The results of this
is to create a reliable method for accurately identifying and study may help create automated and more dependable
segmenting a wide range of cell types or structures in these systems for evaluating blood smear images, which would
large images. The study will most likely concentrate on the benefit medical.
unique challenges of handling such extensive and complex
material, emphasizing the need of feature extraction in the B. Machine Learning Based Approach
image processing pipeline and capturing essential
components such cell shapes, colors, and textures. Apart The study [14] counts and identifies blood cells
from its technological aspects, the project emphasizes the automatically using a machine learning technique. They
significance of image processing for biological applications, automatically count and identify blood cells using the Yolo
particularly for precise semantic segmentation of blood cell (You Only Look Once) algorithm, with 96.09% accuracy for
images to support medical research and diagnostics. RBCs and 86.89% accuracy for WBCs. This is based on a
examination of 364 annotated 100x magnified photos from
The project's main focus is on the research [7] use of image
the Blood Cell Count Dataset (BCCD). The YOLO model is
processing algorithms to identify anemia from microscopy
modified with three outputs and threshold levels defined by
blood smear pictures. The main objective of the research is calculating the average absolute error. While this approach is
to create an image processing system that can identify useful in avoiding incorrect diagnoses, it can occasionally
anemia, a disorder marked by a deficiency of hemoglobin or lead to multiple platelets counts. To tackle this issue, they
red blood cells. It entails examining microscope blood employ the K-nearest neighbour (KNN) and intersection
smear pictures, which offer comprehensive details on the over union (IOU) techniques. The study [15] focuses on the
application of machine learning to the prediction of anemia, Some vision tasks where deep learning approaches have
a major medical concern. This study focuses on a number of demonstrated outstanding performance include object
important areas related to machine learning, including the tracking, image categorization, and pattern recognition.
prediction of anemia, the use of pertinent datasets for testing Rather of having human engineers develop the feature
and training, the choice of suitable machine learning extraction approach from data, DL technology is built on the
algorithms, feature engineering to find pertinent health idea that it may be taught using a general-purpose learning
indicators, training models with historical data, evaluating methodology. Convolutional neural networks (CNNs) have
the efficacy and accuracy of the models, and potential shown to be amazing image processors in the deep learning
clinical uses of the machine learning model in
space. Constructing a comprehensive model to utilize a
mammography and early anemia diagnosis. In addition to
CNN-driven workflow is not overly challenging. CNN
highlighting the value of technology in healthcare, this study
shows how machine learning can help anticipate and detect based deep learning architecture also allows the avoidance
disorders like anemia early on, which could result in major of laborious hand-crafted feature design while retaining the
improvements in patient care and diagnostics. The study [6] required performance. This led to the development of CNN-
offers a comprehensive analysis of the application of based techniques for diagnosing anemia. The research [1]
machine learning methods to the classification of Deep learning algorithms were used to automate the
microscopic images of blood cells, especially in the context counting and differentiation of blood cell types in a blood
of computer-aided diagnostics. The many machine learning sample. Using a collection of annotated photos of blood
techniques for image classification—a crucial step in illness samples, the project will build and train a deep learning
detection and medical diagnostics—are thoroughly assessed model, such as a convolutional neural network (CNN). Deep
and summarized in this review. A discussion of numerous learning enables semantic segmentation, which enables the
machine learning techniques and algorithms, including model to identify and classify each pixel in an image in
decision trees, support vector machines, neural networks, and order to differentiate between platelets, RBCs, and WBCs.
others, is probably going to be included. The datasets and The primary goal of the project is to improve the accuracy
sources of blood cell pictures used to develop and evaluate and efficiency of blood cell counting, which could lead to
machine learning models are anticipated to be covered in the significant developments in medical research and diagnosis.
review. The goal of the study [9] is to identify blood The main technique used in this work is deep learning,
disorders using conventional machine learning methods. which attempts to automate the process of identifying and
Developing a machine learning-based system that can
counting various blood cell types in an entire blood sample
recognize different blood illnesses like anemia, leukaemia,
[3]. Using deep neural networks, namely convolutional
and other haematological conditions is the main objective of
this research. Several traditional machine learning methods, neural networks (CNNs) or other advanced architectures,
including k-nearest neighbours, decision trees, and support this program creates accurate and efficient models for the
vector machines, will be examined in this study in order to detection of red blood cells (RBCs), platelets, and white
analyse pertinent haematological and medical data for the blood cells (WBCs). These algorithms have been trained on
purpose of disease identification. Most commonly, feature a set of tagged photos of blood cells, enabling them to
engineering is employed to extract valuable information comprehend the intricate features and patterns specific to
from the datasets used to train and test these machine each type of cell. Deep learning classifies individual cells
learning algorithms. Machine learning models are useful for down to the pixel level, enabling semantic segmentation.
diagnosing and classifying diseases, helping to identify The examination [16] Deep learning models were developed
certain blood abnormalities. The creation of a dual-objective with an incredible 99.54% accuracy for the diagnosis of
machine learning model was the main emphasis of the sickle cell anemia (SCA) in the research paper "Deep
research [18]. Its goal is to create a model for calculating learning models for classification of red blood cells in
blood hemoglobin levels, a crucial metric in the diagnosis of microscopy images to aid in sickle cell anemia diagnosis,"
anemia and other haematological disorders. Secondly, it written by Alzubaidi et al. The researchers used data from
involves classifying anemia according to predicted erythrocytes IDB, ALL-IDB, and other online sources to
hemoglobin levels and potentially additional clinical develop three Convolutional Neural Network (CNN) models
information. Although precise project specifics are not given, with different layers and filters. These models' extracted
it is anticipated that machine learning, regression, and
characteristics were used to train a multi-class Support
classification techniques would be utilized to forecast
Vector Machine (SVM), which produced an accuracy range
hemoglobin levels and divide people into several anemia
groups. The study [11] provides a comprehensive of 98–99%. Serious arterial blockages with potentially lethal
examination of the ways in which conventional machine outcomes are characteristic features of sickle cell disease
learning techniques are used to categorize white blood cells (SCD), also known as sickle cell anemia.
(WBCs) in blood smear pictures. It investigates the use of The study [5] presents a novel deep learning technique for
models such as support vector machines, decision trees, and automatically identifying and counting cells in images of
k-nearest neighbours, offering insights into their suitability microscopic blood. Deep learning, a type of machine
for WBC classification. In the area of medical diagnostics, learning that creates neural network models that can
these methods are used for feature extraction, dataset recognize intricate patterns in photos, is the basis of this
selection, and image preprocessing. These traditional investigation. The authors provide a comprehensive
machine learning models are expected to be subjected to description of the development and training of their deep
performance comparisons and evaluations, with a focus on learning model, emphasizing its superior performance over
measures such as accuracy and sensitivity. other methods that result in more precise and efficient cell
C. Deep Learning-Based Approach counts. The empirical results presented in the research show
how successful this deep learning approach is and how it has
the potential to revolutionize medical picture processing.
This study highlights the deeper benefits of deep learning in
the medical domain by streamlining labor-intensive
processes that were previously done by hand.
The study [8] Deep learning is being used to automate the
identification and separation of red and white blood cells in
tiny blood images. The central technology in this study is
deep learning, which is a subfield of machine learning.
Complex image patterns are analyzed using neural network
topologies. The authors present a region proposal technique
that identifies regions of interest in images that most likely III. METHODOLOGY
contain blood cells by applying deep learning algorithms. Blood smear analysis is a crucial method for detecting
Only this automated method can provide accurate cell anemia, a condition characterized by a deficiency in the
quantification, which is crucial for medical diagnosis and number or quality of red blood cells (RBCs). The
research. This work is expected to shed light on the deep methodology for detecting anemia involves several key steps
learning model's architecture, training procedure, and use of aimed at analysing blood samples obtained from a digital
datasets, highlighting the model's ability to produce accurate microscope. Initially, the process involves preprocessing,
differentiation. The research [10] Deep learning is needed to where blood samples are collected and prepared for analysis
automatically identify and detect sickle cells in blood using a digital microscope. This step ensures that the samples
samples. It is expected that the paper outlining this novel are adequately prepared and free from artifacts that could
method that makes use of advanced deep learning will go interfere with accurate analysis.
into great detail about the deep learning model's
architecture, training set, and methodology, all of which
contribute to the machine's amazing accuracy in identifying
these various blood cells. The remarkable ability of deep
learning to recognize intricate patterns and characteristics in
photos results in precise and effective sickle cell
identification. techniques for training the system to
recognize and distinguish between healthy and sickle cells
in images of microscopic organisms. The research [12]
presents a novel method that combines Deep Neural
Networks (DNNs) and Locality Sensitive Hashing (LSH) to
create artificial blood smears. In order to create synthetic
smears, LSH is employed to effectively locate comparable
patches from real blood smear images. DNNs are used to
further refine these artificial images in order to improve
realism and faithfully mimic the features of actual blood
smears. This approach aims to solve the lack of labeled data
in medical image analysis, especially in hematology, for
training machine learning models. The efficiency of the
suggested method in producing artificial blood smears that
closely mimic real ones is demonstrated by experimental Fig 1 : Block Diagram of Proposed Method
results. These artificial images can provide useful training
data for machine learning algorithms targeted at different
hematological tasks, like segmentation and classification. The next critical aspect of the methodology involves
Potential uses for the technique include medical research, classification techniques. Three primary approaches are
teaching, and the creation of diagnostic instruments. It also employed for anemia classification: image processing,
offers a way to enhance already-existing datasets, which machine learning, and deep learning algorithms. Image
makes it easier to design and assess blood smear analysis processing techniques are utilized to enhance and analyze the
algorithms. In general, the suggested methodology signifies images obtained from the blood samples. Machine learning
a propitious addition to the domain of medical picture algorithms are employed to identify patterns and
synthesis and analysis, capitalizing on progressions in relationships within the extracted features, aiding in the
hashing methodologies and deep learning architectures. classification of different types and severities of anemia.
Deep learning algorithms, known for their ability to process
complex data and identify intricate patterns, play a pivotal
role in improving the accuracy and efficiency of anemia
classification.
In conclusion, the methodology for detecting anemia through
blood smear analysis involves preprocessing of blood
samples, feature extraction based on RBC characteristics,
and the application of various classification techniques,
including image processing, machine learning, and deep
learning algorithms. This comprehensive approach enables
accurate and efficient detection and classification of anemia,
facilitating timely intervention and management of this
prevalent blood disorder.
Preprocessing: The focus shifts to feature extraction. In this
step, the structure of RBCs is analyzed and classified based
on three primary features: the size, color, and shape of the
RBCs. These features provide valuable insights into the
morphology and characteristics of the red blood cells, which
are indicative of various types of anemia
Feature extraction: The dataset comprises 1421 samples
featuring six attributes: gender, hemoglobin levels, mean
corpuscular hemoglobin (MCH), mean corpuscular
hemoglobin concentration (MCHC), mean corpuscular Fig 3. Distribution of hemoglobin levels by gender
volume (MCV), and result. These attributes collectively
provide insights into various hematological parameters.
Through comprehensive analysis, researchers can discern
patterns and correlations within the dataset, potentially
aiding in the diagnosis and understanding of hematological
conditions and their implications across different genders and
hemoglobin levels.
IV. RESULT

Fig 4. Confusion Matrix

Reference Precision Recall F1-Score

Lee et.al 70.1 86.7 77.51

[2]

Chy et.al 90 94.3 92.09

[3]
Fig 2: Distribution of hemoglobin levels by anemia results
Jaiswal 86.3 91.3 85.94
A machine learning code has been used to analyze et.al [5]
hemoglobin and mean corpuscular volume (MCV) data,
yielding important insights into the identification of anemia
and its subtypes in both men and women. The specified Hortinela 83.3 87.5 85.34
ranges of hemoglobin count for men (12–15 gm/dL) and et.al [17]
women (11.5–14.5 gm/dL) served as important reference
points when diagnosing anemia. Notably, people who had Alsheref 94.61 97.59 96.07
hemoglobin levels above or within the designated range et.al [13]
were classified as non-anemic, whilst people whose levels
were below the threshold were classified as anemic. The
inclusion of MCV measurements (80.0-95.0fL for men and Proposed 98.9 100 99.5
82.0-98.0fL for women) further refined the classification of Method
anemia. By examining the MCV values, it was possible to
distinguish between micro lytic and macro lytic anemia.
Instances where MCV fell below the established range were A confusion matrix is a tabular representation that provides
indicative of micro lytic anemia, whereas values exceeding an overview of a classification model's performance by
the range pointed towards macro lytic anemia. Where the fig contrasting the anticipated and true labels. Fig 4 shows how
2 shows that the distribution of hemoglobin levels by many of the model's predictions were true positives (TP),
anemia results.
true negatives (TN), false positives (FP), and false negatives
(FN).
 [7] Nithya, R., and K. Nirmala. "Detection of Anemia using Image
Processing Techniques from microscopy blood smear images." In
Journal of Physics: Conference Series, vol. 2318, no. 1, p. 012043.
IOP Publishing, 2022.
[8] Di Ruberto, Cecilia, Andrea Loddo, and Lorenzo Putzu. "Detection of
red and white blood cells from microscopic blood images using a
region proposal approach." Computers in biology and medicine 116
(2020): 103530.
[9] Alsheref, Fahad Kamal, and Wael Hassan Gomaa. "Blood diseases
detection using classical machine learning algorithms." International
Journal of Advanced Computer Science and Applications 10, no. 7
(2019).
[10] de Haan, Kevin, Hatice Ceylan Koydemir, Yair Rivenson, Derek
Tseng, Elizabeth Van Dyne, Lissette Bakic, Doruk Karinca et al.
"Automated screening of sickle cells using a smartphone-based
microscope and deep learning." NPJ digital medicine 3, no. 1 (2020):
76.
[11] S. Grampurohit and C. Sagarnal, "Disease Prediction using Machine
Learning Algorithms," 2020 International Conference for Emerging
Technology (INCET), Belgaum, India,2020, pp.1-7,
doi:10.1109/INCET49848.2020. 9154130..
Fig 5. Parameter Analysis [12] Al-Qudah, Rabiah, and Ching Y. Suen. "Synthetic blood smears
generation using locality sensitive hashing and deep neural
V. CONCLUSION networks." IEEE Access 8 (2020): 102530-102539.
An effective strategy for addressing the worldwide health [13] Safuan, Syadia Nabilah Mohd, Mohd Razali Md Tomari, and Wan
Nurshazwani Wan Zakaria. "White blood cell (WBC) counting
issue of anemia is the automated image processing method analysis in blood smear images using various color segmentation
that has been suggested. This project intends to give medical methods." Measurement 116 (2018): 543-555.
professionals a dependable, affordable, and non-invasive [14] Alam, Mohammad Mahmudul, and Mohammad TariqulIslam.
tool for the early identification and classification of anemia "Machine learning approach of automatic identification and counting
based on red blood cell morphological abnormalities by of blood cells." Healthcare technology letters 6, no. 4 (2019): 103-108
utilizing advances in digital imaging and machine learning. [15] Jaiswal, Manish, Anima Srivastava, and Tanveer J. Siddiqui. "Machine
learning algorithms for anemia disease prediction." In Recent Trends
The algorithm's predictive and diagnostic precision are in Communication, Computing, and Electronics: Select Proceedings
further improved by adding clinical data. If this strategy is of IC3E 2018, pp. 463-469. Springer Singapore, 2019.
successfully put into practice, patient outcomes might be [16] Alzubaidi, Laith, Mohammed A. Fadhel, Omran AlShamma, Jinglan
greatly enhanced and the burden of anemia-related problems Zhang, and Ye Duan. "Deep learning models for classification of red
blood cells in microscopy images to aid in sickle cell anemia
on global healthcare systems could be reduced. This project diagnosis." Electronics 9,no. 3 (2020): 427.
demonstrates how interdisciplinary partnerships in data [17] Rahadi, Irwan, Meechoke Choodoung, and Arunsri Choodoung. "Red
science, technology, and medicine may address difficult blood cells and white blood cellsdetection by image processing." In
health issues and enhance public health outcomes. Journal of Physics: Conference Series, vol. 1539, no. 1, p. 012025.
[18] El-kenawy, E. S. M. T., and E. SM. "A Machine Learning Model for
REFERENCES Hemoglobin Estimation and Anemia Classification." International
Journal of Computer Science and Information Security (IJCSIS) 17,
no. 2 (2019): 100- 108.
[1] HemaSri, Allaparthi, Mopuru Devi Sreenidhi, Valluru Venkata Krishna
Chaitanya, Gamidi Vasanth, V. Murali Mohan, and T. Satish. [19] Hortinela, Carlos C., Jessie R. Balbin, Janette C. Fausto, Paul Daniel
"Detection of RBCs, WBCs, PlateletsCount in Blood Sample by using C. Divina, and John Philip T. Felices. "Identification of abnormal red
Deep Learning." In 2023 International Conference on Sustainable blood cells and diagnosing specific types of anemia using image
Computing and Data Communication Systems (ICSCDS), pp. 47-51. processing and support vector machine." In 2019 IEEE 11th
IEEE, 2023. International Conference on Humanoid, Nanotechnology, Information
Technology, Communication and Control, Environment,
[2] Chy, Tajkia Saima, and Mohammad Anisur Rahaman. "A comparative andManagement (HNICEM), pp. 1-6. IEEE, 2019
analysis by KNN, SVM & ELM classification to detect sickle cell
anemia." In 2019 International conference on robotics, electrical and
signal processing techniques (ICREST), pp. 455-459. IEEE, 2019.
[3] Lee, Shin-Jye, Pei-Yun Chen, and Jeng-Wei Lin. "Complete blood cell
detection and counting based on deep neural networks." Applied
Sciences 12, no. 16 (2022): 8140.
[4] Shahzad, Muhammad, Arif Iqbal Umar, Muazzam A. Khan, Syed
Hamad Shirazi, Zakir Khan, and Waqas Yousaf. "Robust method for
semantic segmentation of whole-slide blood cell microscopic
images." Computational and mathematical methods in medicine 2020
(2020).
[5] Di Ruberto, Cecilia, Andrea Loddo, and Lorenzo Putzu. "A region
proposal approach for cells detection and counting from microscopic
blood images." In Image Analysis and Processing–ICIAP 2019: 20th
International Conference, Trento, Italy, September 9–13, 2019,
Proceedings, Part II 20, pp. 47-58. Springer International Publishing,
2019.
[6] AL-DULAIMI, K. H. A. M. A. E. L., and Teba Makki. "Blood Cell
Microscopic Image Classification in Computer Aided Diagnosis
Using Machine Learning: " Iraqi Journal for Computer Science and
Mathematics 4, no. 2 (2023): 43-55.