1
CELL IMAGE ANALYSIS FOR DETECTION
OF MALARIA
 line 1: 2nd Given Name Surname
line 1: 1st Given Name Surname
line 2: dept. name of organization
(of Affiliation)
line 3: name of organization
(of Affiliation)
line 4: City, Country
line 5: email address or ORCID
line 1: 4th Given Name Surname
line 2: dept. name of organization
(of Affiliation)
line 3: name of organization
(of Affiliation)
line 4: City, Country
line 5: email address or ORCID

line 1: 3rd Given Name Surname
line 2: dept. name of organization line 2: dept. name of organization
(of Affiliation) (of Affiliation)
line 3: name of organization line 3: name of organization
(of Affiliation) (of Affiliation)
line 4: City, Country line 4: City, Country
line 5: email address or ORCID line 5: email address or ORCID
line 1: 5th Given Name Surname
line 2: dept. name of organization
(of Affiliation)
line 3: name of organization
(of Affiliation)
line 4: City, Country
line 5: email address or ORCID
 2

Abstract—Malaria is an infectious disease that causes over

400,000 deaths per year. Malaria is a true endemic in some areas
of the world, meaning that the disease is regularly found in the
region. When assessing the risk of infectious disease outbreak we
typically examine how many people in the population or at or
below poverty levels. Regions of the world that are below poverty
levels most likely do not have access to proper healthcare. We will
use cell images and differentiate between the infected cell and the
normal one’s using deep learning.
In this work we will detect malaria using cell images. We
capitalize on recent developments of deep learning and propose a
novel algorithm based on a Convolutional Neural Network
(CNN). In neural networks, Convolutional neural network
(ConvNets or CNNs) is one of the main categories to do images
recognition, images classifications. Objects detections,
recognition faces etc., are some of the areas where CNNs are
widely used.

Not surprisingly, an area of the world that either has a corrupt

I. INTRODUCTION
As shown in Fig 1.1 malaria remains a major burden on global
health, with roughly 200 million cases worldwide and more
than 4,00,000 deaths per year. Malaria is caused by protozoan
parasites of the genus Plasmodium that are transmitted
through the bites of infected female Anopheles mosquitoes
and that infect the red blood cells.
So, what makes some areas of the world more susceptible to
malaria while others are totally malaria free?
A. POVERTY LEVEL
When assessing the risk of infectious disease outbreak we
typically examine how many people in the population or at or
FIG 1.1: Malaria death rate per 100,000 population
government or is experiencing civil war willalso
have higher poverty levels and lower access to proper
healthcare.
Furthermore, if may be impossible for a corrupt government to
provide emergency medical treatment or issue proper
quarantines during a massive outbreak.
D. DISEASE TRANSMISSION VECTORS
A disease vector is an agent that carries the disease and
spreads it to other organisms. Mosquitoes are notorious for
carrying malaria.
Once infected, a human can also be a vector and can spread
malaria through blood transfusions, organ transplants, sharing
needles/syringes etc.
Furthermore, warmer climates of the world allow mosquitoes
to flourish, further spreading disease.

below poverty levels.The higher the poverty level, the higher

Without proper healthcare, these infectious diseases can lead
the risk of infectious disease.
to endemic proportions.

B. ACCESS TO PROPER HEALTHCARE

Why we have chosen this project?
Regions of the world that are below poverty levels most likely
do not have access to proper healthcare.
Without good healthcare, proper treatment, and if necessary,
quarantine, infectious diseases can spread quickly.
C. WAR AND GOVERNMENT
 Is the area war-torn?
 Malaria have been doing a great loss in the region of
Africa and we have the cure of it but early detection
of malaria is something that is significant in order to
take care of it but that is not possible in the region of
Africa as we have discussed above.
 We need a low cost infrastructure to start our project.

 Setup is very easy.

 Is the government corrupt?

II. LITERATURE REVIEW

 Is there in-fighting mongst the states or regions of a
country? A. TECHNOLOGY
In our project we have used CNN (Convolutional Neural
Network). CNN is now the go-to model on every image
 3

There are a lot of terms being used so let’s visualize them one
by one.
FIG 2.2: Input is converted to feature map.

On the left side is the input to the convolution layer, for

example the input image. On the right is the convolution filter,
also called the kernel, we will use these terms interchangeably.
This is called a 3x3 convolution due to the shape of the filter.

We perform the convolution operation by sliding this filter over

related problem. The main advantage of CNN compared to its
predecessors is that it automatically detects the important
features without any human supervision. For example, given
many pictures of cats and dogs it learns distinctive features for
each class by itself.
the input. At every location, we do element-wise matrix
multiplication and sum the result (Fig 2.3). This sum goes into
the feature map. The green area where the convolution
operation takes place is called the receptive field. Due to the
size of the filter the receptive field is also 3x3.
FIG 2.3: Element-Wise Matrix Multiplication
Here the filter is at the top left, the output of the convolution
operation “4” is shown in the resulting feature map. We then
slide the filter to the right and perform the same operation,
adding that result to the feature map as well.

CNN is also computationally efficient. It uses special We continue like this and aggregate the convolution results in
convolution and pooling operations and performs parameter the feature map (Fig 2.4).
sharing. This enables CNN models to run on any device,
making them universally attractive.

B. ARCHITECTURE OF CNN
All CNN models follow a similar architecture (Fig 2.1). There
is an input image that we’re working with. We perform a
series
FIG 2.1: CNN architecture

convolution + pooling operations, followed by a number of

fully connected layers. If we are performing multiclass
classification the output is softmax.

C. CONVOLUTION
The main building block of CNN is the convolutional layer.
FIG 2.4: Input to feature map using filter
Convolution is a mathematical operation to merge two sets of
Information. In our case the convolution is applied on the input
data using convolution filter to produce a feature map (fig 2.2).

D. POOLING
After a convolution operation we usually perform pooling to
reduce the dimensionality. This enables us to reduce the
number of parameters, which both shortens the training time
and combats overfitting. Pooling layers downsample each
feature map independently, reducing the height and width,
keeping the depth intact.
The most common type of pooling is max pooling which just
takes the max value in the pooling window (Fig 2.5). Contrary
to the convolution operation, pooling has no parameters. It
slides a window over its input, and simply takes the max value
E. LATEST PROGRESS
F.
FIG 2.5: Pooling window
in the window. Similar to a convolution, we specify the
window size and stride.
Here is the result of max pooling using a 2x2 window and
stride 2. Each color denotes a different window. Since both the
window size and stride are 2, the windows are not overlapping.
Note that this window and stride configuration halves the size
of the feature map. This is the main use case of pooling,
downsampling the feature map while keeping the important
information
Now let’s work out the feature map dimensions before and
after pooling. If the input to the pooling layer has the
dimensionality 32x32x10, using the same pooling parameters
described above, the result will be a 16x16x10 feature map.
Both the height and width of the feature map are halved, but
the depth doesn’t change because pooling works independently
on each depth slice the input.
By halving the height and the width, we reduced the number of
FIG 2.6: Reduce number of weights
weights to 1/4 of the input (Fig 2.6). Considering that we
typically deal with millions of weights in CNN architectures,
this reduction is a pretty big deal.
DISADVANTAGE
4
 If the input image is blurry it is difficult to do the
prediction on that image and chances are that it will
show an incorrect prerdiction.
 Time for training the dataset is higher.
 We large dataset to be able to do the prediction.
 We cannot directly give the input to the model project
as there is a manual intervention is required.
Machines have been built up so as to directly capture the cell
image from the blood samples.
KEY HIGHLIGHTS
 Disease prediction in probability.
 Dataset is verified as it was released by the US
government medical department.
 This algorithm was been used previously for leaf
disease prediction, we have converted it for malaria
detection.
 The time taken by this algorithm to run on current
dataset has been reduced from 24 to 20 hours.
 Processing is done without using GPU.
III. PROBLEM STATEMENT
There are a handful of methods to test for malaria:
 Blood smears
 Antigen testing
The blood smear process can be visualized by:
 First, a blood sample is taken from a patient and then
placed on a slide.
 The sample is stained with a contrasting agent to help
highlight malaria parasites in red blood cells.
 A clinician then examines the slide under a
microscope and manually counts the number of red
blood cells that are infected.
In order to help make malaria testing a faster process in the
field, scientists and researchers have developed antigen tests
for Rapid Diagnosis Testing (RDT).
An ideal solution would, therefore, need to combine
the speed of RDTs with the accuracy of microscopy.

IV. METHODOLOGY
The objective is to tell with reasonable certainty whether a
patient has malaria or not. To do so following steps will be
taken
 Train the neural network by using the malaria image
dataset.
 Mapping the erythrocytes in the blood smear which
can be further used for malaria detection.
 Obtain microscopic image of the blood sample of
patient.
 Input the image to the neural network.
 Report the result to the patient whether he/she has
malaria.
V. RELATED WORK
In the first paper[1] , highly sensitive optical technique for
detection of blood cells infected by malaria by using third
harmonic generation imaging of hemozoin pigment which gets
successively deposited in the parasite during its life. The THG
method enables to detect malaria robustly and with high
degree of accuracy. The accuracy was found to be 95% for
cells infected by THG. Though the process is fast and accurate
but expensive hence cannot be used for normal malaria
detection .
Next paper [2] is used for study the concept of convulation
neural network(CNN) for image analysis to diagonose
malaria. . This study proposes a novel machine learning model
based on convolutional neural network (CNN) to classify
images blood smear as affected with malaria or unaffected.
ten-fold cross-validation was used based on 27,577 single cell
images, the accuracy ofthe 16-layer CNN model was found
to be 95%. In the paper non-microscopic (blood) images were
used.
Now we study about the Evaluations of Deep Convolutional
Neural Networks for Automatic Identification of Malaria
Infected Cells [3] . This paper studied automatic identification
of malaria infected cells using deep learning methods. In this
whole slide images of thin blood stains are used to compile a
dataset of malaria-infected red blood cells and non-infected
cells, as labeled by a group of four pathologists. Simulation
results showed that all these deep convolution neural networks
achieved classification accuracies of over 95%, higher than the
accuracy of about 92% attainable by using the support vector
machine method.
For more focusing images we study the CNN based Malaria
Diagnosis from Focus-stack of Blood Smear Images Acquired
using Custom-built Slide Scanner[4] .This paper introduces a
focus stacking‐based approach for automated quantitative
detection of Plasmodium falciparum malaria from blood
smear. The proposed approach of employing sophisticated
algorithmic processing together with inexpensive
instrumentation can potentially benefit clinicians to enable
malaria diagnosis. Since the slide scanner is of low cost but
5
this process is costlier than other process of detection of
malaria infected cells.
Tomari, R., Zakaria, W. N. W et al. study Artificial Neural
Networks for Detection of Malaria and propose use of
Artificial Neural Networks (ANN) for the diagnosis .The
features / parameters needed for dataset were computed from
the data obtained by the holographic images of the blood cells
and is given as input to ANN which then classifies the image
as affected or unaffected with malaria. Using this technique an
accuracy up to 90% was obtained.
In this paper [6] we study, Clinical malaria diagnosis:
rule-based classification statistical prototype, In this study, we
identified predictors of malaria, developed data mining,
statistically enhanced rule-based classification to diagnose
malaria and developed an automated system to incorporate the
rules and statistical models, yet its treatment remains very
expensive for the majority of the patients to afford. The
overall sensitivity of the rule-based classification obtained
from the predictive dataset was 70 % [68–74; 95 % CI] with a
specificity of 58 % [54–66; 95 % CI]. The values for both
sensitivity and specificity varied by age, generally showing
better performance for the data mining classification rules for
the adult patients.
In this paper [7] we study, Plasmodium vivax classification
from digitalization microscopic thick blood film using
combination of second order statistical feature extraction and
K-Nearest Neighbor (K-NN) classifier method, we propose an
accurate method to classify plasmodium vivax from
digitalization microscopic thick blood film using combination
of second order statistic feature extraction and K-Nearest
Neighbor (K-NN) classifier method. In this feature extraction,
we use GLCM (Gray Level Co-occurrence Matrix) to get
contrasts, correlations, energys, and homogeneity values.
Based on the result of experiments, the combination of second
order statistical and K-NN has a high accuracy for classifying
plasmodium vivax with average accuracy 95%.
In this paper [8] we study, automated malaria parasite and
there stage detection in microscopic blood images, the
following work aims an automated system which will detect
the malaria parasite along with its stage in which it is. Various
Image processing techniques are used in this proposed
method. In first level of classification, the presence of malaria
parasite in the Red Blood Corpuscles will be checked and then
the stage and type of parasite will be determined using multi-
stage Support Vector Machine. The expected results will be
around 95%.
In this paper [9] we study, Segmentation of erythrocytes
infected with malaria parasites for the diagnosis using
microscopy imaging, in this article, an edge-based
segmentation of erythrocytes infected with malaria parasites
using microscopic images has been developed to facilitate the
diagnostic process. Fuzzy C-means clustering is applied to
extract infected erythrocytes. The experimental results showed
 6

that the proposed method can gain 98%, 93.3%, 98.65% and
90.33% of sensitivity, specificity, prediction value positive and
prediction value negative, respectively.

In this paper [10] we study, Feature extraction and

classification for detection malaria parasites in thin blood
smear, this paper is developed based on the image processing
technique to detect three stages of Plasmodium parasites while
in human host, i.e. trophozoite, schizont, and gametocyte
plasmodium falciparum. Multilayer perceptron
backpropagation algorithm is used to classify all features. The
results show that the proposed method achieves accuracy of
87.8%.

In this paper [11], we study malaria severity classification

through jordan–elman neural network based on features
extracted from thick blood smear, in this the proposed
classifier is developed based on the Jordan–Elman neural
networks. The effectiveness of the classifier is compared to a
support vector machine and multiple regression models.
Reliability of the proposed model achieves 90%. The Jordan–
Elman neural networks classifier can assist medical
practitioners in the speedy detection of malaria and
determining its severity. This will eventually reduce the rate of
morbidity, premature births, and maternal and infant mortality.

Mahdieh Poostchi, Kamolrat Silamut et al.[12] gave an

overview of the techniques and discussed the current
developments in image analysis and machine learning for
microscopic malaria diagnosis and further organized the
different approaches published in the literature according to
the techniques used for imaging, image preprocessing, parasite
detection and cell segmentation, feature computation, and
automatic cell classification. Riemer Sorgedrager [13]
proposed a framework using a convolutional neural network
as a pixel classifier to localize the erythrocytes. Afterwards
classified it accordingly, using a convolutional neural network
as an object classifier. Algorithm successfully localized the
erythrocytes with an average sensitivity of 97.31% and
precision of 92.21%. Jonathan M. Be´lisle, Santiago
Costantino et al. [14] we reported a novel and highly sensitive
optical-based detection of malaria-infected blood cells by third
harmonic generation (THG) imaging of hemozoin pigment
that is naturally deposited by the parasite during its lifecycle.
The THG signal from the hemozoin was greater than we have
observed in any cell type with signal/noise ratios that reach
1000:1. Christine F. Mark Walter et al. [15] proposed a method
of Magnetic bead based simultaneous ELISA & Sequential
detection (SCSD) for detection of malaria biomarkers – pLDH
and PfHRPII. Purnima Pundit et al. [16] analyzed the Digital
Holographic Interferometric Microscopic (DHM) image for
detection of infected erythrocytes using Artificial Neural
Networks technology. Hanung Adi Nugroho et al. [17]
examined the classification of malaria parasite and proposed a
method for detection of malaria parasite stages using K-means
clustering and Multilayer Perceptron Neural Networks.

VI. FLOW CHAR VII. ALGORITHM

shuffle images ( path p)

{
set random seed 42
shuffle images
 define set of callbacks
fit model F
} 7
predict data(validation generator Vg, testing generator Tg2,
VIII. RESULTS model F)
{
Positive Negative resnet testing generation
Positive 4796 160 calculate probability pb
Negative 161 4809 find index of each images using largest pb
show report
}

IX. CONCLUSION

Using microscopic images CNN has been successfully applied

to detect malaria with accuracy of 97%. With precision, recall,
f-measure of 96.75%, 96.77% and 96.75% respectively.
Processing is done in 20 hours using CPU power (Fig 8.1).

TABLE 8.1: Confusion Matrix

true positives
Recall=
true positives+ false negatives
…(i)

 RECALL = 4796/4796+160

=0.9677

true positives
Precision=
true positives+ false positives
…(ii) FIG 8.1: Epoch vs. Loss/Accuracy

 PRECISION = 4796/4796+161

=0.9675

2∗precision∗recall
F 1= …
precision+recall
(iii)

 F1 = 2*(0.9675*0.9677)/(0.9675+0.9677)

= 0.9675

tp+ tn
Accuracy = …
tp+tn+fp +fn
(iv)

 ACCURACY = 0.97
 8

[7] Rahmanti, F. Z., Ningrum, N. K., Imania, N. K., &

Purnomo, M. H. (2015, November). Plasmodium vivax
classification from digitalization microscopic thick blood film
using combination of second order statistical feature extraction
and K-Nearest Neighbor (K-NN) classifier method. In 2015
REFERENCES
4th International Conference on Instrumentation,
Communications, Information Technology, and Biomedical
[1] Bélisle, J. M., Costantino, S., Leimanis, M. L., Bellemare,
Engineering (ICICI-BME) (pp. 79-83). IEEE.
M. J., Bohle, D. S., Georges, E., & Wiseman, P. W. (2008).
Sensitive detection of malaria infection by third harmonic
[8] Charpe, K. C., & Bairagi, V. K. (2015, January).
generation imaging. Biophysical journal, 94(4), L26-L28.
Automated malaria parasite and there stage detection in
microscopic blood images. In 2015 IEEE 9th International
[2] Liang, Z., Powell, A., Ersoy, I., Poostchi, M., Silamut, K.,
Conference on Intelligent Systems and Control (ISCO) (pp. 1-
Palaniappan, K., ... & Huang, J. X. (2016, December). CNN-
4). IEEE.
based image analysis for malaria diagnosis. In 2016 IEEE
International Conference on Bioinformatics and Biomedicine
[9] Somasekar, J., & Reddy, B. E. (2015). Segmentation of
(BIBM) (pp. 493-496). IEEE.
erythrocytes infected with malaria parasites for the diagnosis
using microscopy imaging. Computers & Electrical
[3] Dong, Y., Jiang, Z., Shen, H., Pan, W. D., Williams, L. A.,
Engineering, 45, 336-351.
Reddy, V. V., ... & Bryan, A. W. (2017, February). Evaluations
of deep convolutional neural networks for automatic
[10] Nugroho, H. A., Akbar, S. A., & Murhandarwati, E. E. H.
identification of malaria infected cells. In 2017 IEEE EMBS
(2015, October). Feature extraction and classification for
International Conference on Biomedical & Health Informatics
detection malaria parasites in thin blood smear. In 2015 2nd
(BHI) (pp. 101-104). IEEE.
International Conference on Information Technology,
Computer, and Electrical Engineering (ICITACEE) (pp. 197-
[4] Gopakumar, G. P., Swetha, M., Sai Siva, G., & Sai
201). IEEE.
Subrahmanyam, G. R. K. (2018). Convolutional neural
network‐based malaria diagnosis from focus stack of blood
[11] Chiroma, H., Abdul-kareem, S., Ibrahim, U., Ahmad, I.
smear images acquired using custom‐built slide
G., Garba, A., Abubakar, A., ... & Herawan, T. (2015). Malaria
scanner. Journal of biophotonics, 11(3), e201700003.
severity classification through Jordan-Elman neural network
based on features extracted from thick blood smear. Neural
[5] Tomari, R., Zakaria, W. N. W., Jamil, M. M. A., Nor, F. M.,
Network World, 25(5), 565.
& Fuad, N. F. N. (2014). Computer aided system for red blood
cell classification in blood smear image. Procedia Computer
[12] Mahdieh Poostchi, Kamolrat silamut, Richard j. Maude,
Science, 42, 206-213.
Stefan Jaeger, and George Thoma Image analysis and
machine learning for detecting malaria. U.S. National Library
[6] Bbosa, F., Wesonga, R., & Jehopio, P. (2016). Clinical
of Medicine, National Institutes of Health, Bethesda, MD
malaria diagnosis: rule-based classification statistical
20894
prototype. Springerplus, 5(1), 939.
 9

[13]Jonathan M. Be´ lisle, Santiago Costantino, Mara L.

Leimanis, Marie-Jose´ e Bellemare, D. Scott Bohle,
Elias Georges,y and Paul W. Wiseman Sensitive Detection of
Malaria Infection by Third Harmonic Generation Imaging
Department of Physics, McGill University, Montreal, Quebec,
Canada.

[14] Riemer Sorgedrager High sensitive malaria diagnosis

using convolutional neural networks in an on-field setting.
Technical University Delft.

[15] Christine F.Markwalter, KeerstenM.Ricks, AnnaL.Bitting,

LwiindiMudenda, DavidW.Wright. Simultaneous capture and
sequential detection of two malarial biomarkers on magnetic
micro particles. Elsevier - Talanta. 2016 AUG; 161:443–9.

[16] Purnima Pandit, A. Anand. Artificial Neural Networks for

Detection of Malaria in RBCs. 2016 AUG; arXiv: 1608.06627.

[17] Hanung Adi Nugroho, Son Ali Akbar et al. Feature

Extraction and Classification for Detection Malaria Parasites
in Thin Blood Smear. IEEE- 2nd Int. Conference on
Information Technology, Computer and Electrical Engineering
(ICITACEE), Indonesia. 2015 OCT