A Portable System for Screening of Cervical Cancer

Paloma Cepeda Andrade and Sesh Commuri

University of Nevada, Reno, 1664 N Virginia St, Reno, NV, E.U.A.

Keywords: Cervical Cancer, Portable Colposcope, Specular Reflections, Segmentation.

Abstract: Cervical cancer is one of the most common cancers that affect women, with the highest incidence and

mortality rates occurring in low- and middle-income countries. Early detection is crucial for successful

treatment, but the need for expensive equipment, trained colposcopists, and clinical infrastructure has made

it difficult to eradicate this disease. To address such limitations, we propose the development of a portable,

low-cost colposcope that is easy to use, which uses image processing techniques to automate lesion detection

and provides a quantitative measure to evaluate progression of the disease or to measure treatment efficacy.

Through this paper, we present the development of a system that encompasses the above, and preliminary

results show that we can achieve a low-cost bioinformatics-based screening for early detection of cervical

cancer in a clinical setting.

1 INTRODUCTION

Cervical cancer is a preventable and curable disease.

Despite this, it is the fourth most common cancer

amongst women worldwide, with an estimated

604,000 new cases and 342,000 deaths reported in

2020 ((WHO), 2022). It has been reported that around

90% of these cases occurred in low- and middle-

income countries (Sung et al., 2021), where there is

limited access to healthcare, and it is difficult to carry

out screening and treatment for this disease.

Approximately 95% of cases of cervical cancer

are caused by the human papillomavirus (HPV)

(Franco et al., 2001). Prevention of this disease can

be made through vaccination against HPV, as well as

screening methods to identify cervical dysplasia

(abnormal cells on the cervical tissue) prior to the

development of cervical cancer. The golden standard

for cervical screening is Visual Inspection with

Acetic Acid (VIA) during the colposcopic exam, also

known as a colposcopy (Sankaranarayanan et al.,

2003).

A colposcopy requires a speculum to hold the

vaginal walls open and a light source to illuminate the

surface of the cervix during the examination. The

colposcope is then used to visually examine the

cervix and surrounding tissues. The cervix is first

rinsed with saline solution and then stained with 3-

5% acetic acid. Acetic acid causes abnormal proteins

in the epithelium to coagulate and appear white and

opaque when inspected under light. These abnormal

cells on the surface of the cervix are termed as

cervical intraepithelial neoplasia (CIN) and are

usually caused by certain types of HPV. While CIN

is not cancer, it may become cancer and spread to

adjacent normal tissue. The density of acetowhitening

(AW) indicates the grade of the precancerous cells (Li

& Poirson, 2006). CIN can be easily treated, and full

recovery is possible if it is detected early. Depending

on the severity, a diagnosis of CIN1, CIN2, or CIN3

is given, where a mild case is likely to heal on its own,

and moderate to severe cases require immediate

treatment (Castle et al., 2007). However, even for an

expert in the field, it is difficult to diagnose and

investigate the presence of such lesions unless a

biopsy is performed (Kudva & Prasad, 2018). Other

contributing factors to the high mortality rate in low-

and middle-income countries include lack of

adequate medical facilities, high power requirements

for existing colposcopes, social stigma around

women’s healthcare, and the need for multiple visits

to medical clinics, which may be inaccessible,

unpleasant, and time-consuming for some women

((WHO), 2022). Therefore, there is a need for a tool

with screening capabilities that is easy to use and

reliable in communities that lack the necessary

resources.

Automated detection of CIN has been made

possible by several studies conducted to obtain

images of the cervix. Archived digitized cervical

Andrade, P. and Commuri, S.

A Portable System for Screening of Cervical Cancer.

DOI: 10.5220/0011777700003414

In Proceedings of the 16th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2023) - Volume 1: BIODEVICES, pages 72-79

ISBN: 978-989-758-631-6; ISSN: 2184-4305

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

images from screening, taken with a fixed-focus

camera, were used to develop a method based on deep

learning algorithm for automated visual evaluation of

cervical images (Hu et al., 2019). This dataset was

obtained during a study of HPV and cervical cancer

in Costa Rica (Bratti et al., 2004). Li and

coresearchers (Li et al., 2021) collected a large

dataset consisting of colposcopic images collected

from 8,604 patients, which include pathological

information and annotations from expert physicians.

Finally, the Atlas of Colposcopy (Basu &

Sankaranarayanan, 2017) was developed by the

International Agency for Research on Cancer (IARC)

with the intention to provide a freely accessible

learning tool for researchers and medical personnel.

Challenges to the automated detection of CIN

include low quality of the devices used at times,

which impair the image resolution; lighting

conditions, which can make shadows appear,

hindering the ability to find the cervical region of

interest (ROI); distortion of the images due to the

presence of glare or specular reflections (SR) from

the light source; and the appearance of artifacts such

as the speculum and surrounding tissue.

Efforts to overcome these challenges include

classification of cervical dysplasia using machine

learning methods based on image pre-processing and

feature extraction (Asiedu et al., 2019; Bai et al.,

2019; Fernandes et al., 2018). Other researchers have

focused on image pre-processing techniques to detect

and remove SR from the cervical epithelium (Das &

Choudhury, 2017; Lange, 2005; Meslouhi et al.,

2011). While these research teams have achieved

promising results in the field, there is still a need for

complete systems that can rely on effective

techniques for cervical image analysis while in low-

resource settings.

In this paper we describe a portable colposcope

and its use in automated detection of CIN. We present

details about our ongoing pilot study for data

collection and validation of our image analysis

algorithms. Furthermore, we use a technique that can

help remove specular reflections prior to identifying

the presence of lesions and determining their size in

relation to the cervical ROI. Images from the Atlas of

Colposcopy are used to demonstrate the removal of

SR. The steps in detecting and classifying lesions on

the surface of the cervix are presented in the next

section, and the results are discussed in section 3.

2 EQUIPMENT AND METHODS

2.1 System Overview

In this paper, we propose a low-cost and portable tool

for screening and detecting abnormal changes in the

cervix. The Cervitude Imaging System (CIS) (Patent

Appl. 63/121,432, 2020) consists of a digital

microscope used in conjunction with a speculum

during the colposcopic exam and a desktop

application to analyze the captured images. The

digital microscope is capable of 50 to 1000X

magnification and provides illumination to the cervix

via an embedded LED ring. The microscope is placed

at the opening of the speculum, outside the patient's

body, using the setup shown in Figure 1. It is then

powered on and used to capture digital images of the

cervix. The images are then processed using the

proprietary CIS algorithm to detect the presence, size,

and extent of lesions on the cervix.

Figure 1: CIS Digital Microscope with Speculum.

The CIS desktop application, developed in

MATLAB, is built around an image analysis

algorithm that can quickly detect and locate lesions

on the surface of the cervix as images are collected.

Figure 2 shows our framework for image acquisition

and analysis. After capturing images of the cervix

with our probe, the images are transferred via

Bluetooth to a laptop where the desktop application is

running. The CIS application pre-processes the image

in three stages: removal of specular reflections;

segmentation of the cervical region of interest; and

segmentation of acetowhite (AW) lesions.

Upon completing the image pre-processing steps,

classification to determine different severities of CIN

will be performed. In this paper, we cover the steps

involved in the removal of SR. Identification of ROI

can be found in (Cepeda-Andrade & Commuri,

2022).

A Portable System for Screening of Cervical Cancer

Figure 2: Steps for Lesion Detection and Classification.

2.2 Pilot Study

We obtained approval from the Institutional Review

Board (IRB number 1629150-4) from the University

of Nevada, Reno to conduct a pilot study to acquire

cervical images with CIS. This small-scale study,

consisting of 30 participants, is ongoing, and we

expect to finalize our image acquisition step in the

upcoming weeks.

The study is directed towards adult female

participants aged between 21 and 65 years old that

have been identified with having abnormal cervical

screening, and where follow-up evaluation is

recommended. Informed consent is required from all

patients participating in this study. The process of

consent is as follows: upon checking in for their

appointment, a member of the research team hands

the patient an information form explaining the details

of the research and any preliminary questions the

patient might have about the study will be answered

by the principal investigator. If the patient is

interested in participating in the study, she will go

through the consent form with the physician who will

be performing her colposcopy.

Following our protocol, standard procedure for

colposcopy is followed, that is, the speculum is

placed, the cervix visualized, the CIS probe is then

placed in a clean protective sleeve and inserted into

the speculum and an image is captured using the CIS

image capturing system and colposcope. Next, 5%

acetic acid (AA) solution is applied to the cervix and

the cervix will be evaluated by the standard

colposcope. The findings are documented by the

standard colposcope and the CIS system. Directed

biopsy and ECC (endocervical curettage) is obtained

as indicated utilizing the standard colposcope.

To protect the privacy of research participants,

once the initial organization of data is completed, the

name of the participant will be replaced by her initials

and a five-digit identification number. In addition to

collecting images of the cervix, we also use the CIS

desktop application to collect demographic

information and relevant health information as

directed by the participating physicians. A

registration form (shown in Figure 3) is filled out for

every new patient. There are also options to attach the

camera and capture images or video, as well as an

option to quickly analyze and save these images. All

data except for identifying information is collected

through this application and is then encrypted and

stored in a secure server.

2.3 Removal of Specular Reflections

(SR)

As mentioned earlier, abnormal cells present in the

epithelium of the cervix become opaque when treated

with acetic acid. The reflected light from the opaque

epithelium gives it a white color, indicating the

presence of a precancerous lesion.

Specular reflections are observed when light from

the colposcope is reflected from the cervical

epithelium or from the speculum. SR on the cervical

images will also appear white, but with high

brightness and low saturation values. This could

cause a problem in which reflections of light on the

tissue could be misconstrued as AW lesions,

producing incorrect diagnostic results. Therefore, it is

necessary to identify and remove SR areas without

affecting the acetowhite regions in the image.

We approach the removal of specular reflections

through exemplar-based image inpainting (Criminisi

et al., 2004; Le Meur et al., 2013). It is important to

maintain the structure and texture of the surrounding

epithelial tissue or AW lesion, and this method has

been shown to take into account these factors to

properly restore images (Shroff & Bombaywala,

2019).

The method to achieve this begins with

identifying the target regions, Ω, to be removed and

inpainted. Given a patch, Ψ

, centered on a boundary

pixel, p, in the target region, its priority can be

computed as:

𝑃

(

𝐩

)

=𝐶

(

𝐩

)

𝐷

(

𝐩

)

(1)

where 𝐶

(

𝐩

)

is a confidence term that measures

the amount of reliable information surrounding the

pixel p and 𝐷

(

𝐩

)

is a data term that reflects the

presence of contour information. These are defined

as:

𝐶

(

𝐩

)

∑

𝐶(𝐪)

𝐪∈



∩

Ψ

𝐩



(2)

BIODEVICES 2023 - 16th International Conference on Biomedical Electronics and Devices

𝐷

(

𝐩

)

∇𝐼

𝐩



⋅𝐧

𝐩



𝛼

(3)

where |Ψ

| is the area of Ψ

, 𝛼 is a normalization

factor (𝛼 = 255 for an image of type uint8), ∇I

┴

shows the direction of curves of constant light

intensity on a surface (isophotes), and n

is a unit

vector orthogonal to the point p. The priority term is

calculated for every border patch, with different

patches for each pixel on the boundary of the target

region.

Next, data is extracted from the source region, Φ,

which is computed by subtracting the target region

from the input image. Image texture is propagated by

directly sampling the source region until a patch,

centered around a point q, most similar to Ψ

is found:

𝒒



=argmin𝑑(Ψ

𝐩

,Ψ

𝐪

)

(4)

where the distance between the two patches is

calculated through the sum of squared differences

(SSD). Finally, after finding the exemplar Ψ

, the

pixel values are copied into the target patch and the

confidence values are updated.

A summary of the described algorithm for

removal of specular reflections is given below:

1. Identify target regions, Ω, from the input

image, to be removed and inpainted.

2. Create a binary mask where the nonzero

values are the pixels that correspond to the

target regions.

3. Identify the source region, Φ, which

corresponds to the input image minus the

target region.

4. For every patch of size Ψ centered on a

boundary pixel in the target region, compute

the patch priority.

5. Find the patch with the maximum priority.

This patch constitutes the target patch to be

inpainted.

6. Given the target patch, search for the best-

matching patch in the source region by using

the sum of square difference (SSD).

7. Copy image data from the best-matching

patch to the target patch.

8. Update the input image, binary mask, and

patch priority.

Steps 4–8 are then repeated until all target regions

have been inpainted. We applied this algorithm to

regions identified as specular reflections in our

dataset, and the results are shown in section 3.

2.4 Segmentation of the Cervical

Region

We presented the steps to achieve segmentation of the

cervical region in detail in (Cepeda-Andrade &

Commuri, 2022). We follow a five-step approach to

the analysis of cervigrams and detection of

precancerous lesions on the cervix, summarized as

follows:

Figure 3: Example of the CIS desktop application.

A Portable System for Screening of Cervical Cancer

1. Convert the cervical image from sRGB to

LAB color space and combine the

information from the L* and a* channels.

2. Use the k-means algorithm (Luo et al., 2003;

Tariq & Burney, 2014) to obtain clusters,

segment the image, and identify the cervical

ROI.

3. Implement morphological filters (Burger &

Burge, 2016) to eliminate holes and connect

similar regions.

4. Automatically crop the segmented and

filtered image to maximize the ROI.

5. Identify AW lesions and calculate their area

in proportion to the cervix.

2.5 Annotation Tool

Through the pilot study, we seek to assess the

Cervitude Imaging System’s functionality and ease of

use at each step of colposcopy exam compared to a

standard colposcope while observing the patient’s

comfort level throughout the evaluation. The quality

of screening information and accuracy of detection

will also be evaluated.

For this purpose, we have developed our own

image annotation tool. The participating physicians

have agreed to help with annotating the images that

we collect. They can select between segmenting the

cervical ROI, lesions on the cervical epithelium, or

specular reflections. Their input is extremely valuable

to our ongoing research, as we will pursue

quantitative, rather than qualitative, assessments

regarding the performance of our segmentation and

lesion detection algorithms. Their input also helps us

as we increase the size of our dataset through further

clinical studies. A large data set is necessary to

increase the accuracy of automated detection through

machine learning techniques.

Figure shows an example of the information that

can be collected from cervigrams. The segmented

regions are saved as binary masks and masked images

to facilitate further analysis by the research team.

3 RESULTS AND DISCUSSION

3.1 Removal of Specular Reflections

We implemented the algorithm described in section

2.3 on the Atlas of Colposcopy dataset. A sample of

results is presented in Figure 4.

Figure 4: Left: Original image. Right: Results of SR

removal.

In the first three rows of the above figure, we

show images that do not present signs of

precancerous lesions. Without the removal of SR,

these images would likely be classified as containing

abnormalities, prompting the medical provider

conducting the examination to perform a biopsy in the

“white” areas for further analysis. The post-processed

images present a more accurate state of the epithelial

tissue, which reduces the possibility of a misdiagnosis

of misclassification by an automated system.

The algorithm also works when SR is present on

top of a lesion, as seen on the bottom row of Figure 4.

The reflection of light is no longer obstructing the

details on the lesion, while the texture and structure

matches the surrounding area. This enables the

medical provider to more easily understand where the

acetowhitening is occurring and decide what the best

treatment option is.

A limitation to this algorithm is that it will not

work properly if the target regions are relatively large.

BIODEVICES 2023 - 16th International Conference on Biomedical Electronics and Devices

Figure 5: Segmentation options through our annotation tool. From left to right: cervical ROI, AW lesions, and SR.

As seen on the third row of Figure 5, the texture

remaining after removing the SR does not blend very

smoothly with the surrounding tissue. This may affect

analysis in situations where it is crucial to examine

morphological characteristics such as tissue shape,

mosaics, and punctuation vessels, which are also used

for determining a diagnosis. Further work is being

conducted to improve the observed results.

3.2 Segmentation of the Cervical

Region

Figure 6 shows the region of interest (ROI)

segmentation results on the Atlas of Colposcopy

dataset. The first column shows the original RGB

colposcopy image. The second column displays the

resulting binary mask outline after our segmentation

algorithm. The third column represents the segmented

colposcopic image without implementing the SR

removal algorithm. We add a fourth column

displaying segmentation results after removing SR.

To evaluate the performance of the SR removal

algorithm when automatically segmenting the

cervical region, we compare the area of the ROI when

removing SR and when SR are not removed. Table 1

shows the area of the cervical ROI of the images

presented in Figure 6 when performing our

segmentation algorithm for each of these cases.

By removing areas with specular reflections, the

area of the ROI becomes smaller. A visual assessment

also indicates that in most cases, the accuracy of ROI

segmentation increased after removing SR. Two clear

examples of this are in rows 2 and 4 of Figure. When

minimizing sources of error, such as reflection of

light on the speculum and cervix, as well as

surrounding tissue, it is possible to increase the

accuracy of detection of abnormalities that may lead

to cervical cancer.

This is a promising step towards automation of

AW segmentation and classification of precancerous

lesions.

Figure 6: ROI Segmentation Results. Left to right: Original

image; binary mask; segmented image; segmented image

after SR removal.

Table 1: ROI Segmentation Results.

Row

(Fi

ure)

ROI area with

SR (%)

ROI area

without SR (%)

1 51.89 50.38

2 58.82 49.88

3 64.02 57.37

4 53.32 42.28

5 71.17 64.42

6 50.60 49.28

A Portable System for Screening of Cervical Cancer

4 CONCLUSIONS

In this paper, we propose a portable, low-cost, reliable

system for automatic detection of precancerous

lesions on the cervix. We described our ongoing pilot

study, where we seek to assess the functionality and

reliability of the Cervitude Imaging System (CIS) and

validate our image analysis algorithms. By

developing an application where physicians can

collect relevant information about their patients, as

well as storing images from each visit within the same

location, we facilitate the process around screening

for cervical cancer.

We used a technique that helps remove specular

reflections as the first step in our image pre-

processing procedure. Through this algorithm, we can

remove specular reflections around and within areas

in the cervix that show precancerous lesions. It is an

important step, given that it is important to not only

detect signs of abnormal cells in the cervix, but also

to reduce misdiagnosis and unnecessary biopsies.

Removing specular reflections also improves the

results of segmentation of the cervical region of

interest. Therefore, our image pre-processing method

further decreases the chances of incorrect diagnosis

and treatment.

Future work includes implementing our methods

to images that we collect through our pilot study.

Extensive analysis to increase the accuracy of CIS

will be performed to our images as we increase the

size of our dataset. We believe that our low-cost

bioinformatics-based tool addresses the challenges to

cervical cancer screening in areas where there is

limited access to technology and trained specialists.

ACKNOWLEDGEMENTS

The authors would like to thank Dr. Charles Johnson,

MD, and Dr. Alison Westfall, MD, for their

participation and support in this pilot study.

REFERENCES

(WHO), W. H. O. (2022). Cervical Cancer. Retrieved May

12, 2022 from https://www.who.int/news-room/fact-

sheets/detail/cervical-cancer

Asiedu, M. N., Simhal, A., Chaudhary, U., Mueller, J. L.,

Lam, C. T., Schmitt, J. W., . . . Ramanujam, N. (2019).

Development of Algorithms for Automated Detection

of Cervical Pre-Cancers With a Low-Cost, Point-of-

Care, Pocket Colposcope. IEEE Trans Biomed Eng,

66(8), 2306-2318. https://doi.org/10.1109/TBME.2018

.2887208

Bai, B., Du, Y., Li, P., & Yuchun, L. (2019). Cervical

Lesion Detection Net. 2019 IEEE 13th International

Conference on Anti-counterfeiting, Security, and

Identification (ASID),

Basu, P., & Sankaranarayanan, R. (2017). Atlas of

Colposcopy – Principles and Practice. IARC

CancerBase. https://screening.iarc.fr/atlascolpo.php

Bratti, M. C., Rodríguez, A. C., Schiffman, M., Hildesheim,

A., Morales, J., Alfaro, M., . . . Herrero, R. (2004).

Description of a seven-year prospective study of human

papillomavirus infection and cervical neoplasia among

10000 women in Guanacaste, Costa Rica, . Rev Panam

Salud Publica, 15(2), 75-89. https://doi.org/10.1590/

s1020-49892004000200002

Burger, W., & Burge, M. J. (2016). Digital Image

Processing (2 ed.). Springer-Verlag London.

https://doi.org/10.1007/978-1-4471-6684-9

Castle, P. E., Stoler, M. H., Solomon, D., & Schiffman, M.

(2007). The relationship of community biopsy-

diagnosed cervical intraepithelial neoplasia grade 2 to

the quality control pathology-reviewed diagnoses: an

ALTS report. Am J Clin Pathol, 127(5), 805-815.

https://doi.org/10.1309/PT3PNC1QL2F4D2VL

Cepeda-Andrade, P., & Commuri, S. (2022). Automatic

Segmentation of the Cervical Region in Colposcopic

Images. 15th International Joint Conference on

Biomedical Engineering Systems and Technologies -

BIODEVICES,

Criminisi, A., Pérez, P., & Toyama, K. (2004). Region

filling and object removal by exemplar-based image

inpainting. IEEE Trans Image Process, 13(9), 1200-

1212. https://doi.org/10.1109/tip.2004.833105

Das, A., & Choudhury, A. (2017). A novel humanitarian

technology for early detection of cervical neoplasia:

ROI extraction and SR detection. 2017 IEEE Region 10

Humanitarian Technology Conference (R10-HTC),

Dhaka.

Fernandes, K., Cardoso, J. S., & Fernandes, J. (2018).

Automated Methods for the Decision Support of

Cervical Cancer Screening Using Digital Colposcopies.

IEEE Access, 6, 33910-33927. https://doi.org/

10.1109/ACCESS.2018.2839338

Franco, E. L., Duarte-Franco, E., & Ferenczy, A. (2001).

Cervical cancer: epidemiology, prevention and the role

of human papillomavirus infection. CMAJ, 164(7),

1017-1025.

Hu, L., Bell, D., Antani, S., Xue, Z., Yu, K., Horning, M.

P., . . . Schiffman, M. (2019). An Observational Study

of Deep Learning and Automated Evaluation of

Cervical Images for Cancer Screening. J Natl Cancer

Inst, 111(9), 923-932. https://doi.org/10.1093/jnci/

djy225

Kudva, V., & Prasad, K. (2018). Pattern Classification of

Images from Acetic Acid-Based Cervical Cancer

Screening: A Review. Crit Rev Biomed Eng, 46(2),

117-133. https://doi.org/10.1615/CritRevBiomedEng.

2018026017

BIODEVICES 2023 - 16th International Conference on Biomedical Electronics and Devices

Lange, H. (2005). Automatic glare removal in reflectance

imagery of the uterine cervix. Proceedings of SPIE -

The International Society for Optical Engineering,

Le Meur, O., Ebdelli, M., & Guillemot, C. (2013).

Hierarchical super-resolution-based inpainting. IEEE

Trans Image Process, 22(10), 3779-3790.

https://doi.org/10.1109/TIP.2013.2261308

Li, W., & Poirson, A. (2006). Detection and

Characterization of Abnormal Vascular Patterns in

Automated Cervical Image Analysis. In G. Bebis, R.

Boyle, B. Parvin, D. Koracin, P. Remagnino, A. Nefian,

G. Meenakshisundaram, V. Pascucci, J. Zara, J.

Molineros, H. Theisel, & T. Malzbender, Advances in

Visual Computing Berlin, Heidelberg.

Li, Y., Liu, Z. H., Xue, P., Chen, J., Ma, K., Qian, T., . . .

Qiao, Y. L. (2021). GRAND: A large-scale dataset and

benchmark for cervical intraepithelial Neoplasia

grading with fine-grained lesion description. Med

Image Anal, 70, 102006. https://doi.org/10.1016/

j.media.2021.102006

Luo, M., Ma, Y.-F., & Zhang, H.-J. (2003). A Spatial

Constrained K-means A pproach to Image

Segmentation Joint Conference of the

Fourth International Conference on Information,

Communications and Signal Processing, 2003

and Fourth Pacific Rim Conference

on Multimedia Singapore.

Meslouhi, O., Kardouchi, M., Allali, H., Gadi, T., &

Benkaddour, Y. (2011). Automatic detection and

inpainting of specular reflections for colposcopic

images. Open Computer Science, 1(3).

Sankaranarayanan, R., Wesley, R., Thara, S., Dhakad, N.,

Chandralekha, B., Sebastian, P., . . . Nair, M. K. (2003).

Test characteristics of visual inspection with 4% acetic

acid (VIA) and Lugol's iodine (VILI) in cervical cancer

screening in Kerala, India. Int J Cancer, 106(3), 404-

408. https://doi.org/10.1002/ijc.11245

Shroff, M., & Bombaywala, M. (2019). A qualitative study

of Exemplar based Image Inpainting. SN Applied

Sciences, 1(12). https://doi.org/https://doi.org/

10.1007/s42452-019-1775-7

Sung, H., Ferlay, J., Siegel, R. L., Laversanne, M.,

Soerjomataram, I., Jemal, A., & Bray, F. (2021). Global

Cancer Statistics 2020: GLOBOCAN Estimates of

Incidence and Mortality Worldwide for 36 Cancers in

185 Countries. CA Cancer J Clin, 71(3), 209-249.

https://doi.org/10.3322/caac.21660

Tariq, H., & Burney, S. M. A. (2014). K-Means

Cluster Analysis for Image Segmentation.

International Journal of Computer Applications, 96.

A Portable System for Screening of Cervical Cancer