International Research Journal of Engineering and Management Studies (IRJEMS)

Volume: 03 Issue: 04 | April -2019 e-ISSN: 2395-0126

Non–Invasive Detection of Anemia Using MATLAB

Mohammad Yusuf1, Kinnera Vinay Kumar2, K.S.S.Rohith3, Jallepalli Naga Raju4

M.R.N Tagore5
1U.G. Student, Department of ECE, VVIT, Nambur, Guntur, Andhra Pradesh, India

2U.G. Student, Department of ECE, VVIT, Nambur, Guntur, Andhra Pradesh, India

3U.G. Student, Department of ECE, VVIT, Nambur, Guntur, Andhra Pradesh, India

4U.G. Student, Department of ECE, VVIT, Nambur, Guntur, Andhra Pradesh, India

5Professor, Department of ECE, VVIT, Nambur, Guntur, Andhra Pradesh, India

--------------------------------------------------------------------***---------------------------------------------------------------------

Abstract-World Health Organization (WHO) identifies 1. INTRODUCTION

Anemia as maligning a quarter of the total world
population, which is considered to be a health hazard In between the years 1995 and 2005, according to a
study conducted by the world health organization
condition marked by the deficiency of Red blood cells or
(WHO), 24.8 percentage of the entire world population
hemoglobin in the blood stream. Hence, an automated, was considered to be anemic [1]. The Hemoglobin
quick, and reliable detection of anemia is essential. Initial concentration in the human blood is considered as the
detection of anemia is actually undertaken visually by the gold standard for the detection of anemia. This is an
physician after examining the color of the anterior intravenous process which needs specialized surgical
conjunctiva of the eye and later confirming it with an equipment. Recently, finger prick blood sample is taken
for lab testing. Higher chances, that it could affect the
invasive blood test. In our study, we have designed a
workers with blood borne infections [2]. Normally, many
technique for the automated detection of anemia through clinics examine the conjunctival pallor of the eyerapidly
non-invasive visual method. In this, the main operation of screen for anemia.Physicians generally pull down the
the detection of anemia is by analyzing the anterior eyelid and subjectively examine the color of the anterior
conjunctival pallor of the eye. It works by quantifying the conjunctival pallor membrane. The clinical sign for
conjunctival color from digital photographs of the eye anemia detection can prove to be quite useful in many
taken with a smartphone camera of appropriate cases, but still the lack of inter-observer agreements in
many situations and low sensitivity of anterior
resolution under sufficient lighting conditions. These
conjunctival color can undermine the authenticity of the
images are then further processed to attain the red and visual detection process [3]. To alleviate the problem of
green component spectra of the conjunctiva color and inter-observer disagreement and human error to make
compared against a threshold to determine whether the the visual detection process more reliable, color scale
patient is anemic or not. Our study was aimed towards the cards, which consist of the color spectrum and the
automation of healthcare facilities in underdeveloped corresponding hemoglobin concentration is used in
parts of the world lacking proper healthcare facilities like many occasions [4].
hospitals and healthcare centers. Thus, we have developed
Section 2 of this paper is about the existing methods for
a computerized, noninvasive, simple, cost effective, easy to the detection of Anemia. Section 3 explains about our
use, and portable primary screening test for anemia which Non-Invasive method of detecting Anemia. The results
can provide a viable alternative to invasive methods of and the discussion regarding that are given in Section 4.
anemia detection and have major humane impact in the The Conclusion is provided in Section 5.
underdeveloped areas of the world.

Key Words : Detection of Anemia, image, photograph

eye, conjunctival pallor, hemoglobin concentration, non-
invasive method.

© 2019, IRJEMS | www.irjems.com | Page 1

 International Research Journal of Engineering and Management Studies (IRJEMS)
Volume: 03 Issue: 04 | April -2019 e-ISSN: 2395-0126

2. EXISTING METHODS 3.1 Taking the photograph

The primary component which contributes to the We first take a picture of the anterior conjunctiva of the
pigmentation found in human blood is Hemoglobin. It eye in adequate lighting, with a standard back facing
has a bias in reflecting the red component of the light camera of a smartphone. We pull the lower eyelid softly
falling on its surface compared to the green component with the thumb and take the photograph so that the
which it primarily takes in. This is the key reason for the
anterior conjunctiva is in focus and as magnified as
deep reddish appearance of hemoglobin [5]. Hence, by
comparing the red and green components of the RGB possible. The image has considerable effects according
color spectrum of the conjunctival pallor, it is the quality of the picture. Improper picture may affect
determined that a person is either Anemic or Non- the whole process. So, the picture should be taken with
Anemic. Many people are vulnerable to anemia, due to the best camera available. The photo should be taken
lack of proper healthcare and medical facilities in with stable hands as excessive shaking could blur the
underdeveloped countries. This situation can be image and may result in erroneous readings. Also taking
alleviated if an indication of anemia can be estimated
without involving expensive blood tests, which are the picture in daylight is recommended.
unavailable in many of these areas. Even the availability
of doctors or medical workers is inconsistent. Detecting
the presence of anemia in a patient, using non-invasive
methods, which does not include expensive tests or even 3.2 Image Processing using MATLAB
the presence of a doctor or medical worker is of great
help. The presence of anemia in a person can be an The image that is captured is uploaded to MATLAB (Fig.
indication of other diseases like jaundice and lack of 2), and then converted to gray scale (Fig. 3). imfreehand
nutrition. The detection of anemia can serve to indicate tool, in MATLAB image processing toolbox, is used to
the presence of other diseases as well. select the required region using the cursor (Fig. 4). The
outer part of the selected image is converted to black
3. METHODOLOGY (Fig. 5). The original image is now compared to this gray
scale image (Fig. 5) and this gray scale part is cropped in
A flow chart constituting the entire process is presented
the original image (Fig. 6).
in Fig. 1

Fig- 2: Color Image

Fig - 1: A flowchart depicting the major parts of our
algorithm

© 2019, IRJEMS | www.irjems.com | Page 2

 International Research Journal of Engineering and Management Studies (IRJEMS)
Volume: 03 Issue: 04 | April -2019 e-ISSN: 2395-0126

Fig- 3: Gray Scale Image Fig- 4: Gray Scale Image with selected part

Fig- 5: Cropped Gray Scale Image Fig- 6: Cropped Color Image

3.3 Detection of Anemia using RGB 4. RESULTS AND DISCUSSION

Then, extract the RGB color of Anterior Conjunctiva. The algorithm decides the state of a patient by averaging
Next, the red color intensity of the RGB spectrum was the color intensity of the red and green spectrum and
compared to the green color intensity to determine then comparing the two values. A suitable threshold is
whether the person was anemic or non-anemic. Then, estimated and if the difference between the two
the mean red pixel intensities and green pixel intensities spectrums is greater than the threshold value then the
are calculated. Compared to non-anemic patients, we algorithm labels that patient as Non-Anemic. The
found that the difference between the two means was difference is calculated as:
small for anemic patients. To determine the threshold
value, a number of eye images of a subject whose anemia Mean red color intensity – Mean green color intensity
level was previously known, were analyzed. Based on
So in case of moderately anemic patients, the value of the
inspection of data for anemic and non-anemic patient we
difference can also come out to be negative. This
manually selected a threshold difference of means. We
happens when the mean red color intensity is less than
found that a mean difference of 1.5 was suitable for our
the green color intensity.
given data.
The deficiency of red blood cells or Anemia still persists
to be one of the major causes of health hazard in many
underdeveloped regions throughout the world [1]. The

© 2019, IRJEMS | www.irjems.com | Page 3

 International Research Journal of Engineering and Management Studies (IRJEMS)
Volume: 03 Issue: 04 | April -2019 e-ISSN: 2395-0126

first step towards mitigating the risks of anemia is the smartphone with suitable regulation in sufficient
early detection of this syndrome. Most conventional lighting. The image taken is next transferred to a
methods of anemia detection rely on a chemical blood computer by any means.RGB spectrum of the anterior
test performed on the intravenously acquired blood conjunctival pallor is processed bya computer program
sample of the patient in question. However, due to the that processes the image to extract and compares it with
lack of proper medical facilities or hospitals in many a pre-determined threshold value to conclude whether
areas, this intravenous method of anemia detection the subject is anemic or not. The value of the threshold is
becomes highly inconvenient at certain times. In many determined based on our collected data so the quantity
patients, the detection of anemia is eventually delayed of the data has a considerable impact on the accuracy of
until it causes irreversible organ damage which may the system and a larger and more diversified collection
even lead to death in certain occasions. of data injected into our algorithm would substantially
boost the accuracy. Furthermore, the threshold of
Our method of the detection of anemia requires no anemic and non-anemic patients can also depend on
intravenous blood samples and can be quite his/her geographical location which can be incorporated
conveniently administered requiring only the availability into the program if data from different regions is
of a camera enabled smart phone which is becoming available. In the less developed regions of the world.This
increasingly available in recent times. The process is method of anemia detection can prove to be a vital
user friendly with requires no prior specialized skills to method in the fight against anemia. This can result in a
operate making it very convenient to administer. significant increase in the health conditions in those
areas along with a sufficiently less deaths caused by the
The table below shows the patient’s actual anemia state
as per the blood test report, their hemoglobin level and effects of anemia. Moreover, this method can also be
the predictions made by our algorithm. used to detect the lack of blood caused by other means
like blood loss due to physical injury of various organs of
the body.
TABLE 1: List of patients with their actual anemia state
and the predictions made by our algorithm. REFERENCES
Patient Actual Presence of Anemia [1] World Health Organization. “ NutritionalAnaemias.
Number Hemoglobin anemia Prediction Report of a WHO scientific group,” World Heal Organ—
level according to though our
Tech Rep Ser, 405:5-37, 1968.
(gram/100 ml) blood report process
[2] J. Hickner, D.G. Graham, N.C. Elder, E. Brandt, C.B.
1. 8.0 Yes Anemic Emsermann, S.Dovey, “Testing process errors and their
harms and consequences reported from family medicine
2. 11.0 No Non-Anemic practices: a study of the American Academy of Family
3. 10.0 No Anemic Physicians National Research Network,” QualSaf Health
Care. 17: 194–200, 2008
4. 9.3 Yes Anemic [3] J.P. Chalco, L. Huicho, C. Alamo, N.Y. Carreazo, C.A.
Bada, “Accuracy of clinical pallor in the diagnosis of
5. 10.2 No Non-Anemic
anaemia in children: a metaanalysis,” BMC Pediatr. 5: 46,
6. 15.3 No Non-Anemic 2005..
[4] M-E-E.K. Chowdhury, V. Chongsuvivatwong, A.F.
7. 6.8 Yes Anemic
Geater, H.H.Akhter, T. Winn, “Taking a medical history
8. 12.1 No Anemic and using a colour scale during clinical examination of
9. 7.3 Yes Anemic pallor improves detection of anaemia,” Trop Med Int
Health. 7: 133–9, 2002.
10. 12.4 No Non-Anemic [5] M. Setaro, A. Sparavigna, ”Quantification of erythema
using digital camera and computer-based colour image
analysis: a multicentre study,” Skin Res Technol. 8: 84–8,
2002.
5. CONCLUSION

To summarize, we have perfected a non-invasive process

for the successful detection of anemia with 78.9%
accuracy. The process involves taking a picture of the
conjunctival pallor of the eye using the camera of a

© 2019, IRJEMS | www.irjems.com | Page 4