Breast Cancer Detection Using Smart Wearable Devices with Thermal

Sensors

Raniya Ketﬁ

1 a

, Zeina Al Masry

1 b

, Noureddine Zerhouni

1 c

, Catherine Gay

and Christine Devalland

2 d

SUPMICROTECH, CNRS, Institut FEMTO-ST, 24 rue Alain Savary, Besanc¸on, F-25000, France

Service d’Anatomie et Cytologie Pathologiques, H

opital Nord Franche-Comt

e, 100 Route de Moval, 90400 Tr

evenans,

France

Service de Gyn

ecologie Obst

etrique, H

opital Nord Franche-Comt

e, 100 Route de Moval,

90400 Tr

evenans, France

ﬁ

Keywords:

Breast Cancer, Breast Thermography, Breast Cancer Detection, Wearable Devices, Thermal Sensors, Smart

Bra, Early Detection.

Abstract:

Breast cancer is the most frequent cause of cancer-related mortalities among women worldwide. Early detec-

tion of breast cancer is one of the best approaches to prevent this disease. In some developed countries, the

5-year relative survival rate of breast cancer patients is above 90% due to early prevention. Many early detec-

tion tools have been developed and used such as mammography, ultrasounds, and magnetic resonance imaging

(MRI). Still, these tools are not always the best in terms of cost, effectiveness, and risk-free. Developing a

more effective, risk-free, and affordable technique for breast cancer detection has always been a necessity to

increase survivability. Authors have found the potential of non-radiative and non-invasive thermography for

anomaly breast detection. This systematic review aims to provide an introduction and guide for smart wear-

able devices for breast cancer detection using thermal sensors by discussing the advantages of these devices

as well as the challenges of developing and implementing them. A total of 6 relevant works drawn from 286

papers on the subject were carefully analyzed, and the information was synthesized. The selected papers were

synthesized according to the design of the wearable device, its data collection, and classiﬁcation methodolo-

gies. Finally, this review tackles the challenges that come with developing such devices and the great promise

and advantages they hold for early breast cancer detection.

1 INTRODUCTION

Breast cancer is a signiﬁcant health concern world-

wide as it is the most commonly diagnosed can-

cer worldwide (Sung et al., 2021). According to

the World Health Organization, one in eight women

will be diagnosed with breast cancer in their lifetime

(Michaels et al., 2023). Late detection occurs when

the cancerous cells have metastasized and caused dev-

astating results, however, when breast cancer is de-

tected at its early stages, the survival index may go

up to 9O% in high-income countries (Arnold et al.,

2022). Several methods and techniques are used by

https://orcid.org/0009-0003-9224-786X

https://orcid.org/0000-0002-6673-0140

https://orcid.org/0000-0002-8847-3202

https://orcid.org/0000-0002-4128-9264

healthcare to detect breast cancer such as mammog-

raphy, Ultrasound, and Magnetic Resonance Imag-

ing (MRI).While breast cancer screening plays a vi-

tal role in early detection, there are certain limitations

to currently used methods. Mammography requires

compression of the breasts and may cause inconve-

niences to the patient, while exposure to ionizing ra-

diation may even increase the health risk to the patient

(Yaffe and Mainprize, 2011). Dense breast tissue ap-

pears white on mammograms, making it more chal-

lenging to detect abnormalities, as cancerous lesions

can also appear white, women with dense breast tis-

sue may require additional screening methods (Thig-

pen et al., 2018). Ultrasound has its limitations too, as

it may miss smaller masses, resulting in the possibil-

ity of both false-positive and false-negative outcomes

(Halim et al., 2021). Additionally, The quality of ul-

Ketﬁ, R., Al Masry, Z., Zerhouni, N., Gay, C. and Devalland, C.

Breast Cancer Detection Using Smart Wearable Devices with Thermal Sensors.

DOI: 10.5220/0012309400003657

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2024) - Volume 1, pages 23-33

ISBN: 978-989-758-688-0; ISSN: 2184-4305

trasound images can vary depending on the operator’s

skill and experience (Xiao et al., 2015). MRI’s rela-

tively high cost and restricted accessibility can limit

its utilization as a routine screening method. Also,

its high sensitivity may detect noninvasive conditions

that may not progress, potentially leading to a false

diagnosis (Hylton, 2005).

The 5-year relative survival rate in 12 sub-Saharan

African countries stood at 66% between 2008 and

2015 (Sung et al., 2021). Also, the mortality due

to breast cancer is higher in women from poorer

countries and also from lower socioeconomic status

(Tao et al., 2014). Breast cancer detection in un-

derdeveloped or poor countries faces unique chal-

lenges that can limit its effectiveness and accessibil-

ity. These challenges may manifest as restricted ac-

cess to healthcare services, ﬁnancial constraints, and

a lack of awareness regarding breast cancer and its

associated risk factors. These limitations accentuate

the need for a more cost-effective and practical tech-

nique for early breast cancer detection. Many wear-

able devices in the form of bras designed for breast

cancer diagnosis are currently in development and are

at the prototype stage. These devices utilize various

technologies to collect the signal, including thermal

sensors, electrical impedance tomography (EIT), or

ultrasound (Al Masry et al., 2021).

This work explores the emerging ﬁeld of smart

wearable devices designed for breast cancer detec-

tion using thermal sensors. We explore the underly-

ing principles of thermal sensing and its relevance to

breast cancer diagnosis. By highlighting the unique

advantages and challenges of these wearable tech-

nologies, the study aims to provide a comprehen-

sive overview of their potential impact on improv-

ing breast cancer screening methods. This review is

exclusively dedicated to assessing and analyzing sci-

entiﬁc devices aimed at research and diagnostic pur-

poses. We do not cover commercially available prod-

ucts, but instead focus on the technical aspects, per-

formance, and applications of these scientiﬁc instru-

ments.

This paper is organized as follows: Section 2 de-

scribes the methodology used in the review. In section

3 we go through the fundamental principles underly-

ing thermal sensing and its relevance to breast can-

cer detection. Section 4 presents and discusses the

devices developed in the selected papers. Section 5

underlies the advantages and challenges of the stud-

ied wearable devices. Finally, section 6 concludes the

works with future research direction.

2 LITERATURE SEARCH

METHOD

The methodology employed in this study aims to

comprehensively review and synthesize the existing

literature on smart wearable devices for breast can-

cer detection using thermal sensors. Through a sys-

tematic approach, we collected, evaluated, and or-

ganized relevant research articles, conference papers,

and technical reports that contribute to the advance-

ment of this ﬁeld. To identify pertinent sources, we

conducted a rigorous literature search across various

academic databases. A combination of keywords was

employed to ensure the search’s speciﬁcity to our in-

terest: Breast cancer, Breast abnormalities, thermal

sensors, and wearable devices. These keywords are

combined with a search query to get relevant articles

only: (”Breast cancer” OR ”breast anomalies” OR

”breast cancer detection”) AND (”Thermal sensors”

OR (”wearable device” AND ”thermography”)). The

review was conducted on the Google Scholar database

as it includes articles from other speciﬁc databases

such as PUBMED, IEEE, and SCOPUS. The ﬁrst

query returned 286 articles. Excluding the thermal

imaging and thermotherapy keywords with a new

query narrowed it down to 231 articles.

To ensure the selection of high-quality and rele-

vant sources, we established clear inclusion and ex-

clusion criteria:

• Inclusion Criteria: Research papers, conference

papers, and technical reports focusing on wear-

able devices for breast cancer detection using ther-

mal sensors; studies involving thermal sensing

principles, device design, experimental evalua-

tions, and clinical applications were included.

• Exclusion Criteria: Papers published before 2000,

review papers, wearable devices combining other

techniques than thermal sensing, works using

thermal imaging instead of thermal sensing, pa-

pers that do not speciﬁcally address breast can-

cer detection (such as thermotherapy), or non-

wearable systems.

Papers resulting from the query are screened

based on their title, abstract, and keywords, irrelevant

papers are excluded based on the exclusion criteria

cited above. Those screened papers are then read in

full text by the reviewers to assess their contribution

to the topic of interest. A total of 6 papers from 2007

to 2023 were selected to be the most relevant to this

study.

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

3 THERMAL SENSING FOR

BREAST CANCER DETECTION

Breast thermography is a non-invasive technique that

uses infrared cameras or sensors to measure and map

the heat patterns emitted by the breasts (Singh and

Singh, 2020). The underlying principle of breast ther-

mography is based on the fact that abnormal cells,

such as cancer cells, generate more metabolic heat

and alter blood ﬂow patterns in the breast tissue.

Breast thermography is a passive, fast, painless,

moderate, and risk-free imaging technique. It has

been documented that thermography when used with

well-deﬁned protocols, can detect early signs of can-

cer 8 to 10 years earlier than mammography (Singh

and Singh, 2020). Thermography was approved as

an adjunct imaging modality to mammography by the

FDA (Food and Drug Administration) in 1982.

Most of the works found in the literature use ther-

mography with infrared cameras to acquire breast

thermal images. Despite being mostly used, thermal

data acquired by infrared cameras can have many lim-

itations. Some limitations are due to external factors

such as ambient temperature, clothing, or contact with

external heat sources that can impact the surface tem-

perature of the breast and introduce noise or artifacts

in the thermal images. These factors need to be care-

fully controlled or accounted for during data acquisi-

tion to ensure accurate and reliable results. Also, the

quality of thermal data acquired using infrared cam-

eras can be inﬂuenced by the operator’s skill and tech-

nique. Factors such as camera positioning, calibra-

tion, and image capture settings can affect the consis-

tency and reliability of the acquired data.

To take advantage of thermography’s potential to

detect breast cancer without these limitations, some

works propose to acquire the breast thermal matrix

with highly sensitive thermal sensors put in contact

with the skin. Digital thermal sensors may be con-

sidered advantageous for medical applications due to

their ability to provide reliable temperature measure-

ments across a broad range from -55°C to 150°C

(Meijer et al., 2018). Their exceptional accuracy,

which can achieve a precision of 0.01°C within the

human body temperature range, makes them particu-

larly suitable for medical devices. Furthermore, their

compact size, as small as 1mm x 1mm, facilitates in-

tegration into a wide array of medical equipment and

wearable health devices. Additionally, these sensors

are budget-friendly, typically priced at around 4 to

5 US dollars per sensor, enhancing their accessibil-

ity for various healthcare applications. Sensors can

be placed in direct contact with the breast tissue, al-

lowing for more accurate temperature measurements.

This direct contact ensures better thermal coupling

and reduces the potential interference with external

factors.

Sensor-based breast thermography systems may

be more cost-effective compared to infrared cameras

as sensors are generally smaller, more portable, and

less expensive than infrared cameras, making them

more accessible for healthcare facilities or clinics

with limited resources. Combining sensors’ potential

to detect temperature variation very sensitively with a

wearable device such as a brassiere, bra, or wearable

textiles can create a very promising, non-invasive,

cost-effective, and portable detection tool for breast

cancer.

The next section reviews the papers found in the

literature about wearable devices using thermal sen-

sors for breast cancer detection.

4 A COMPARATIVE STUDY

The papers found in the literature are different in

terms of device maturity, some of them are just a

proposition for a wearable bra for breast cancer detec-

tion supported by numerical simulation with COM-

SOL or simulated breast phantoms and others are in

the clinical trial phase for validation.

This comparative study employs a two-step ap-

proach. First, it comprehensively evaluates each

wearable device featured in the selected papers by

a classiﬁcation based on wearable device type, the

number and type of thermal sensors employed, and

the duration of data collection during experiments.

Second, the devices are systematically classiﬁed and

compared with a focus on their respective detection

methodologies. This analysis includes data types,

the size of the testing population, data preprocess-

ing steps, data analysis methodologies, and the per-

formance of the used methodology.

4.1 Wearable Device Design

The selected papers from 2007 to 2023 used very dif-

ferent wearable devices. These devices exhibited dis-

tinct designs, including lightweight wearable patches

integrated with miniature sensors and conventional

textile brassieres with ﬁxed thermal sensors. The

number of sensors employed in these devices varied

signiﬁcantly, ranging from 8 to 28 sensors per breast.

Additionally, the papers employed different types of

temperature sensors with varying accuracies (from

0,75°C to 0.01°C). The collection time also varies

from 30 seconds to 24 hours. Table 1 provides a sum-

mary of the reviewed breast cancer detection devices,

Breast Cancer Detection Using Smart Wearable Devices with Thermal Sensors

offering insights into their diverse design features, the

number and type of sensors used, as well as the test-

ing duration. Additionally, pictures of the devices are

included alongside the table for more clarity and ref-

erence.

The ﬁrst known paper to use thermal sensors to

detect breast cancer in women was (Ng et al., 2007).

They used 8 contact thermal sensors per breast. The

design had more sensors in the upper outer quadrant

of the breast as a high proportion of malignant and

benign diseases arises in this quadrant (Lee, 2005).

Cyrcadia breast monitor (S et al., 2020) is an im-

proved version of the ﬁrst device proposed by (Ng

et al., 2007), more enhancements are added to the

initial version to improve the data capture and analy-

sis processes. The new device is a smart breast patch

equipped with 8 thermal sensors per breast. The ther-

mal sensors used in this device are ADT7420 digital

temperature sensors with a ± 0.25 °C accuracy, which

are compliant with medical device uses as mentioned

in the manufacturer datasheet. The Cyrcadia breast

monitor collects thermal data for 24 hours at every

ﬁve-minute intervals.

In (Laila Fadhillah et al., 2018), the authors also

used a brassiere equipped with the same number of

sensors as (Ng et al., 2007) and (S et al., 2020) (8

sensors per breast). However, it is not known if the

used sensors (LM35 with a ±0,75 °C accuracy) are

adequate for medical or clinical use as it is men-

tioned in their data sheet that they are for power

supplies and battery applications. The authors mea-

sured the temperature simultaneously for a duration

of only 30 seconds, which represents the shortest test

duration among all the devices. In (Antony et al.,

2020), the authors introduced a bra design equipped

with a higher number of sensors per breast compared

to previous devices, ranging from 12 to 20 sensors

per breast, depending on the bra size. These sen-

sors were Nickel Manganate-based NTC chip thermal

probes, which were developed in-house and detailed

in (Arathy et al., ), with an accuracy of ±0.01°C. The

authors in (Elouerghi et al., 2022) employed a ﬂex-

ible card design shaped as a star that incorporated a

signiﬁcantly higher number of sensors compared to

previous studies, their device featured a total of 28

contact thermal mini biosensors per breast. The au-

thors mentioned that the sensors are compliant with

the ASTM E1112 standard (Standard Speciﬁcation

for Electronic Thermometer for Intermittent Determi-

nation of Patient Temperature) and come with an ac-

curacy of 0.1°C. Finally, (Ashreetha et al., 2023) pro-

posed an IOT-based system to collect breast temper-

ature data with a wearable device (a jacket). In this

work, neither the number of sensors nor their type or

the acquisition protocol were speciﬁed.

It is worth mentioning that (S et al., 2020) and

(Antony et al., 2020) are the only papers among the

reviewed studies that introduced variable sizes for

their wearable systems. However, these papers did

not provide extensive details regarding the exact sizes

or the speciﬁc number of sensors allocated to each

size variation of the wearable device.

4.2 Detection Methodologies

As previously mentioned, the reviewed devices dis-

play varying levels of maturity and can be categorized

into two distinct groups. Some devices are primar-

ily focused on demonstrating the feasibility of breast

anomaly detection using contact thermal sensors just

with numerical simulation and physical phantoms

such as (Laila Fadhillah et al., 2018) and (Elouerghi

et al., 2022), while others have progressed beyond this

stage to clinically validate their proposed devices.

These key steps applied for breast cancer detection

for all the reviewed devices are summarized in table

4.2.1 Devices Based On Physical Simulation

In the ﬁrst category, the papers consisted of a proof

of concept for the proposed device. Two of the six

works (Laila Fadhillah et al., 2018) and (Elouerghi

et al., 2022) used a physical phantom and heaters to

mimic the human breast and the tumor in order to col-

lect data. The developed phantoms were very simple.

In (Laila Fadhillah et al., 2018), they used a phan-

tom made of just one layer of agar, while (Elouerghi

et al., 2022) used the same layer and added a 1mm

skin layer made of silicone. Both papers did not men-

tion the thermal properties of the phantom materials

nor highlighted their limitations as they offer a sim-

pliﬁed representation of real breast tissue, lacking the

full complexity and dynamic properties of living tis-

sue. Additionally, the two devices were tested for a

very short period of time (30 seconds for (Laila Fad-

hillah et al., 2018) and one minute according to the

graphs of (Elouerghi et al., 2022)).

In (Laila Fadhillah et al., 2018), authors mea-

sured the temperatures simultaneously for 30 sec-

onds and changed the position of the heater in dif-

ferent quadrants of the phantom. The temperatures

were in the range of 27.34°C to 29.79°C for a nor-

mal phantom while in a heated phantom, the range

was from 30.27°C to 34.18° and higher measurements

were captured in the heated quadrants compared to

the other quadrants which proved that these thermal

sensors are able to detect changes of heat in the phan-

tom. These results are supported by the infrared cam-

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

Table 1: A summary of the reviewed smart wearable devices for breast cancer detection using thermal sensors.

Study Device type Number of sensors

per breast

Type of sensors Time of data

collection

(Ng et al.,

2007)

Sensors connected to

a data recording de-

vice with wires.

8 sensors Not mentioned Not men-

tioned

(Laila Fad-

hillah et al.,

2018)

Wearable Brassiere 8 sensors LM35 sensors with

±0,75 °C accuracy

30 seconds

(S et al.,

2020)

Wearable breast

patch in 6 different

sizes

8 sensors ADT7420 sensors

with ±0.25°C accu-

racy

24 hours

(Antony et al.,

2020)

Stretchable bra with

different sizes

From 12 to 20 sen-

sors depending on

the size of the wear-

able device

Nickel Manganate

based thermal sensor

with ±0.01 °C accu-

racy

30 minutes

(Elouerghi

et al., 2022)

Flexible star shaped

card

28 miniature biosen-

sors

Micro biosensors

with 0.1°C accuracy

Not men-

tioned

(Ashreetha

et al., 2023)

Wearable jacket Not mentioned Not mentioned Not men-

tioned

Breast Cancer Detection Using Smart Wearable Devices with Thermal Sensors

era reading, where the difference between the sensor’s

measurement and the camera’s measurement was in

the range of 0.82°C to 1.27°C.

Similarly, the authors of (Elouerghi et al., 2022)

conducted a comparative analysis against a reference

phantom without heat sources, displaying minimal

temperature ﬂuctuations (∆T < 0.1°C). Another ex-

periment scenario of a tumor located at a depth of

15mm is done, eliciting a temperature disparity of

+0.6°C. (Elouerghi et al., 2022) also compared the

phantom collected data with the numerical simulated

data to ﬁnd a 0.11°C maximal difference. They high-

lighted that their study’s ultimate objective is to ex-

plore alternative sensor options, integrate the gathered

data with artiﬁcial intelligence models, and undergo

clinical validation for their proposed devices.

4.3 Devices Based on Clinical Trials

Within the second category of devices, those undergo-

ing clinical validation, a notable consistency emerges

in the testing procedures. The authors initiated the

process by deﬁning a target population and establish-

ing inclusion and exclusion criteria. After that, they

performed a preprocessing phase to clean and prepare

the collected thermal data for further analysis. This

analysis step involved either the application of tradi-

tional statistical techniques or the training of machine

learning models. The results were evaluated using

metrics such as accuracy, speciﬁcity, and sensitivity.

Data Collection: First, in the data collection step,

authors designated a testing population with speciﬁc

inclusion and exclusion criteria. Inclusion criteria

were similar in all clinical tests, consisting mainly

of age (at least 21 years old), recent breast mammo-

gram availability for healthy subjects, and a biopsy

for patient subjects. Common exclusion criteria were:

pregnant or lactating, previous breast mastectomy or

breast surgery or biopsy within the last 90 days for

healthy subjects. In the study conducted by (Ng et al.,

2007), data was collected from a cohort of 54 indi-

viduals, while (Antony et al., 2020) performed clini-

cal tests involving 60 individuals. Notably, (S et al.,

2020) stood out as the sole study that conducted data

collection at two distinct centers spanning two coun-

tries (Clem Plam Breast Clinic in La Plata, Argentina,

and Ohio State University (OSU) in Ohio, USA).

This multi-center approach to clinical trials holds the

potential to enhance population diversity, ultimately

supporting the validity and generalizability of the pro-

posed device. On the other hand, (Ashreetha et al.,

2023) did not mention any details about the clinical

data collection step, such as the number of partici-

pants or the inclusion and exclusion criteria.

Data Processing: Wearable devices continuously

capture a stream of data, and this data may contain

errors or anomalies due to various reasons, includ-

ing sensor inaccuracies, signal noise, or device mal-

functions. For these reasons, the authors performed

data preprocessing before analyzing or driving con-

clusions from the collected data. In (Ng et al., 2007)

and (S et al., 2020), the authors removed missing data

and outliers while (Ashreetha et al., 2023) replaced

irrelevant and missing data with the mean tempera-

ture which may appear as a better way in order to not

lose valuable information from the collected data. Ac-

knowledging the diversity in individual temperature

proﬁles, authors in (Ng et al., 2007) proceeded to nor-

malize the dataset to a standardized ratio ranging from

0 to 1. Additionally, the authors of (S et al., 2020) and

(Ashreetha et al., 2023) did not use raw temperature

data to detect breast anomalies. (S et al., 2020) used a

wrapper feature selection technique to rank features,

which resulted in using the best 13 features. On the

other hand, (Ashreetha et al., 2023) calculated statisti-

cal features such as Mean, mode, median, range, vari-

ance, and standard deviation in order to use them in a

detection algorithm.

Data Analysis: For analyzing the preprocessed data

in order to detect breast anomalies, the authors used

various detection methodologies. First, (Ng et al.,

2007) used a backpropagation (BPA) neural network

with an input layer, output layer, and 2 hidden lay-

ers and compared it to an RBF-based (Radial Basis

Function) neural network. The RBF model had more

speciﬁc, accurate, and sensitive results compared to

BPA yielding 100% classiﬁcation efﬁciency for nor-

mal and cancer cases, 92% for benign cases, and 90%

for cancer patients. Second, (S et al., 2020) used

the best-extracted features to train several classiﬁers

(Decision Trees, Support Vector Machines, Random

Forest, and Back Propagation Neural Networks. . . ).

The classiﬁer and Best Features combination that pre-

sented the best prediction accuracy are chosen as the

ﬁnal predictive models. It is pertinent to note that

the detailed composition of the extracted features and

the classiﬁer remained undisclosed due to ongoing

patent proceedings. This methodology yielded a pre-

dictive model of considerable performances. This

model demonstrated an accuracy of 78%, sensitivity

of 83.6%, and speciﬁcity of 71.5% under a 10-fold

cross-validation. On the other hand, (Antony et al.,

2020) used a different methodology. (Antony et al.,

2020) author’s work consisted of estimating the tu-

mor’s size and depth and reconstructing a 3D ther-

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

Table 2: Breast cancer detection methodologies.

Study Data Collec-

tion

Population

size

Data preprocessing Data Analysis Results

(Ng et al.,

2007)

Clinical trial 54 individ-

uals

-Removing temperature

from defectuous sensors

and outside the normal

range. - Data normaliza-

tion.

Backpropagation neural

network and Radial basis

function (RBF) classiﬁer

Sensitivity

=91.67%,

Speciﬁcity

=100%, Accu-

racy =92%

(Laila Fad-

hillah et al.,

2018)

physical simu-

lation

None None Studied the difference be-

tween a phantom with a

heater and no heater ∆T

2.93°C < ∆ T

< 4.39°C

(S et al.,

2020)

Clinical trial 93 benign

cases and

108 malig-

nant

-Removing missing values

and outliers. -Best Feature

ranking and selection.

Decision Tree, Support

Vector Machines, Random

Forest, and Back Prop-

agation Neural Network

including bagging and

boosting ensemble tech-

niques.

Sensitivity

=83.6%, Speci-

ﬁcity =71.5%,

Accuracy

=78%

(Antony

et al., 2020)

Numerical and

physical simu-

lation, Clinical

trial.

60 females

(29 patients

and 31

healthy)

None Tumor parameter esti-

mation (location, blood

perfusion, diameter, and

metabolic heat generation)

with FEM and genetic

algorithm. 3D thermal

image construction.

Sensitivity=

82.78%, Speci-

ﬁcity= 87.09%,

Accuracy=85%

(Elouerghi

et al., 2022)

Numerical and

physical simu-

lation

None None Compared between phan-

toms temperatures with and

without heaters

T= 0.1°C with

no heater and

T= 0.6°C with a

heater

(Ashreetha

et al., 2023)

Clinical trials 150 obser-

vations

-Null or irrelevant data is

replaced by the mean tem-

perature. -Mean, mode,

median, range, variance,

and standard deviation of

the breast temperature are

calculated.

Statistical features compar-

ison

Not mentioned

mal image of the breast based on the discrete mea-

sured temperature of the surface of the breast. The pa-

rameter estimation methodology consisted of 3 parts:

forward heat transfer problem, inverse heat transfer

problem, and 3D thermal imaging. The forward heat

transform problem is the breast surface temperature

estimation by a breast numerical model using Penne’s

bioheat equation on the software COMSOL. The in-

verse heat transfer problem aimed to minimize the dif-

ference between experimental and simulation results

using an evolutionary optimization algorithm. The

obtained parameter for these experiments is within an

error of 10% (0.005 W.cm

−3

) for heat generation and

15% (0.3 cm) for tumor size. Also, the proposed esti-

mation methodology yielded a sensitivity of 82.78%

and a speciﬁcity of 87.09% on the clinical data. In

order to differentiate between normal and abnormal

breasts, (Ashreetha et al., 2023) used a conventional

rule-based algorithm to compare the calculated statis-

tical features to show that the asymmetry analysis of

the left and right breasts could differentiate between

abnormal and normal breasts.

4.4 Evaluation Based on Device

Development Process

The reviewed devices yielded very good perfor-

mances in clinical tests, although neither of the stud-

ies tackled an acceptance study before the clinical

tests. An acceptance study is a phase before the clin-

ical trials where the device is evaluated for its ac-

ceptability, feasibility, and practicality among poten-

tial participants and healthcare providers in order to

improve its efﬁciency and integration in the current

healthcare process.

Also, the authors of the reviewed papers presented

the performances of their detection methodologies

without interpretation. The papers mentioned that the

advantage of these wearable devices embedded with

thermal sensors is being able to detect breast abnor-

malities better in mammography, especially in dense

breasts and younger women, but no interpretation of

the used detection models was presented based on the

proportion of dense breasts or age in the studied pop-

ulation. Analyzing the detection efﬁciency based on

Breast Cancer Detection Using Smart Wearable Devices with Thermal Sensors

different categories of breast, age, and ethnicity... can

widely support the validation and the utility of the de-

vice.

The papers did not mention a follow-up clini-

cal investigation to assess the long-term performance,

safety, and clinical utility of the device. A follow-up

study is crucial before validation of this type of clin-

ical device, it helps to explore patient-reported out-

comes, including quality of test, comfort, and satis-

faction with the wearable technology.

Cost-effectiveness is one of the major advantages

that these wearable devices can offer, that’s why as-

sessing the cost of the proposed wearable devices

for breast cancer detection is vital for the effective

integration of this technology into healthcare sys-

tems. Unfortunately, this cost analysis was not tack-

led in any of the reviewed devices, despite its impor-

tance. Understanding the cost structure aids in set-

ting fair pricing strategies and ensuring that patients

have access to these potentially life-saving technolo-

gies. Transparent cost assessments promote account-

ability and help optimize the utilization of healthcare

resources, ultimately facilitating the successful inte-

gration of such innovative devices into clinical prac-

tice while ensuring economic feasibility and patient

accessibility.

In order to compare and evaluate the devices pre-

sented in this section, table 3 summarizes this eval-

uation by checking what has been tackled by the re-

viewed papers and their limitations.

5 DISCUSSION

The studies presented about wearable devices embed-

ded with thermal sensors show the potential of this

new technique for detecting breast cancer in an early

stage. Thermal sensors are proven to be capable of de-

tecting speciﬁc temperature variations that are able to

indicate the presence of breast abnormalities. These

devices hold great promise in the ﬁeld of non-invasive

and cost-effective breast cancer detection. In this sec-

tion, we will present the advantages and challenges

of the mentioned studies. Table 4 summarizes the ad-

vantages, challenges, and areas of improvement for

developing wearable devices embedded with thermal

sensors for breast cancer detection.

One of the most signiﬁcant advantages of these

devices is their ability to detect breast abnormalities at

an early stage. According to (Ng, 2001), it has been

recorded that, with the implementation of carefully

established protocols, thermography has the potential

to identify early signs of cancer approximately 8 to

10 years prior to the detection capabilities of mam-

mography. These devices provide a signiﬁcant beneﬁt

of being non-irradiative as they don’t expose patients

to any radiations of X-rays and are non-invasive by

measuring the temperature only on the breast surface.

Devices embedded with thermal sensors can be exclu-

sively beneﬁcial for breast abnormalities detection in

dense breasts since conventional methods have prob-

lems of false diagnosis in dense breasts, especially

mammography.

These devices can provide a cost-effective breast

cancer detection method. They eliminate the need

for expensive imaging equipment and reduce the ﬁ-

nancial burden on both healthcare systems and pa-

tients. This affordability can make breast cancer

screening more accessible to a broader population,

including those with limited ﬁnancial resources. In

many third-world countries, healthcare resources are

concentrated in urban areas, leaving rural regions un-

derserved. Portable wearable devices can be taken

to remote and rural locations, ensuring that women

in these areas have access to breast cancer screening

without the need for long and costly journeys to urban

centers.

In some communities, discussing breast health

or undergoing breast screening may carry stigma or

taboos. Portable wearable devices can help destigma-

tize these topics by offering a discreet and less inva-

sive way to monitor breast health, potentially encour-

aging more women to participate in screening pro-

grams. While the ﬁeld of wearable devices employing

thermal sensors for breast cancer detection holds im-

mense promise, it is not devoid of challenges. The

pursuit of accurate and reliable detection through this

innovative approach demands a critical examination

of the obstacles that lie ahead. In this discussion, we

unravel the intricacies of these challenges and their

potential impact on the implementation of this trans-

formative technology.

The integration of these devices, with other tech-

nologies holds the potential for advancing breast can-

cer detection in the future. By combining sensors with

artiﬁcial intelligence, machine learning, and cloud

computing techniques, we can improve the accuracy

and effectiveness of diagnosing breast cancer. These

advanced technologies have the capability to analyze

data, identify patterns, and offer valuable insights to

healthcare professionals. Moreover, integrating wear-

able devices with telemedicine platforms can enable

monitoring and consultation, thus increasing access

to breast cancer detection, in underserved regions.

Clinical validation is a critical aspect of the de-

velopment and implementation of wearable medical

devices in healthcare. Conducting rigorous studies

to compare the device’s performance against estab-

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

Table 3: Evaluation of the reviewed studies based on a wearable device development process.

Study Phases

Acceptance Numerical Physical Clinical Multi-centric Data follow-up Cost Commercialization

Study Simulation Simulation Tests Clinical tests Cleaning Study Analysis

(Ng et al., 2007) - - - ✓ - ✓ - - -

(Laila Fadhillah et al., 2018) - - ✓ - - - - - -

(S et al., 2020) - - - ✓ ✓ ✓ - - -

(Antony et al., 2020) - ✓ ✓ ✓ - ✓ - - -

(Elouerghi et al., 2022) - ✓ ✓ - - - - - -

(Ashreetha et al., 2023) - - - ✓ - ✓ - - -

CBRA ✓ ✓ ✓ ✓ ✓ ✓ ✓ - -

Table 4: Advantages, challenges, and areas of improvement for developing wearable devices for breast cancer detection.

Advantages Challenges Areas of improvements

• Early detection of breast Abnor-

malities

• Non-irradiative

• Non-invasive

• Effective in dense breasts

• Affordable and cost-effective

• Accessible for women from

low-income countries

• Wearable, painless, and easy to

use

• Physical and numerical simula-

tions complexity

• The size of the testing popula-

tion

• Patient data privacy

• Clinical trials patients recruit-

ment

• User acceptance and usability

of the device

• Integration in current healthcare

workﬂows

• Enhancing patients recruitment

strategies

• Promoting inclusivity in the

testing population

• Thermal data quality assess-

ment and improvement

• Integrating machine learning

into for analyzing complex ther-

mal patterns to improve the de-

tectio

• Data privacy and security mea-

sures

• Interoapbility with the health-

care system

lished breast cancer diagnostic methods to determine

its sensitivity, speciﬁcity, and overall diagnostic ac-

curacy is not very evident. Clinical trials come with

several challenges that need to be carefully addressed

to ensure the reliability of trial results and the safety

of participants.

First, Finding and enrolling a sufﬁcient number

of eligible participants can be challenging. The re-

cruitment step can be very long and challenging for

clinical trials. Second, achieving a diverse participant

population that represents the broader patient popula-

tion can be difﬁcult. In breast cancer detection clin-

ical trials, the target population must include diverse

age categories, breast type, breast cancer types, breast

size, and even underrepresented groups, such as racial

and ethnic minorities. Achieving population diver-

sity in breast cancer detection clinical trials is essen-

tial for ensuring that research ﬁndings are relevant,

generalizable, and equitable. Efforts to enhance di-

versity should be integrated into the trial design, re-

cruitment strategies, and participant engagement pro-

cesses, with a focus on addressing barriers to partici-

pation and promoting inclusivity in breast cancer re-

search.

In (Laila Fadhillah et al., 2018) and (Elouerghi

et al., 2022)’s work, authors used only physical or

numerical simulation in tests. While these simula-

tions can help collect and assess the device’s per-

formance, but are not enough to validate its use for

diagnostic purposes. Creating accurate breast tissue

models is challenging. Tissue composition can vary

widely between individuals, and accurately represent-

ing this variability in simulations is complex. Sim-

ulations also should replicate the diversity of breast

Breast Cancer Detection Using Smart Wearable Devices with Thermal Sensors

cancer types in size, shape, and location, which is

also very complex due to tumor diversity and inter-

individual variability. Due to these simulation com-

plexities, clinical trials with a sufﬁcient and diverse

population are mandatory to validate wearable de-

vices for breast cancer detection.

Ethical and privacy considerations are of

paramount importance when developing and using

breast wearable devices for breast cancer detection.

These devices collect sensitive health data, and

their usage must adhere to strict ethical and privacy

standards. Obtaining informed consent from users is

crucial, users should fully understand the purpose of

the wearable device, how their data will be collected

and used, and any potential risks or beneﬁts. Breast

wearable devices should employ robust encryption

and data protection measures to safeguard user

information from unauthorized access or breaches.

Ensuring data security is particularly important in the

healthcare context, where data can be sensitive and

personally identiﬁable.

Apart from diagnostic accuracy, clinical valida-

tion should assess the device’s usability in real-world

clinical settings. Factors such as ease of use, inte-

gration into existing healthcare workﬂows, and user

acceptance are important considerations to take in fu-

ture works.

6 CONCLUSION

In conclusion, the ﬁeld of smart wearable devices

equipped with thermal sensors represents a promis-

ing frontier in breast cancer detection. These inno-

vative technologies offer a multitude of advantages,

from non-invasiveness and early detection to acces-

sibility and cost-effectiveness. However, as with any

new technology, there are many challenges to over-

come. Clinical validation, population diversity in tri-

als, ethical considerations, and privacy safeguards are

among the critical issues that demand careful atten-

tion.

Through this review, we can say that smart wear-

able devices with thermal sensors for breast cancer

detection projects are not mature enough to be clini-

cally and widely used, but addressing the challenges

can make these devices more effective, accessible,

and user-friendly. These devices hold the promise

of detecting breast cancer at earlier stages, reducing

healthcare disparities, and transforming breast health

awareness. With continued research, validation, and

collaboration between the medical community and

technology developers, they may well become an ac-

curate and validated breast detection method.

REFERENCES

Al Masry, Z., Zerhouni, N., Gay, C., Meraghni, S., Lodi,

and Devalland, C. (2021). D

etection du cancer du sein

a l’aide de soutiens-gorge connect

es en 2021 : analy-

ses et perspectives. 49(12):907–912.

Antony, L., Arathy, K., Sudarsan, N., Muralidharan, M. N.,

and Ansari, S. (2020). Breast tumor parameter estima-

tion and interactive 3d thermal tomography using dis-

crete thermal sensor data. Biomedical Physics &Engi-

neering Express, 7(1):015013.

Arathy, K., Ansari, S., and Malini, K. A. High reliability

thermistor probes for early detection of breast cancer

using skin contact thermometry with thermal imaging.

10(5):620–628.

Arnold, M., Morgan, E., Rumgay, H., Mafra, A., Singh, D.,

Laversanne, M., Vignat, J., Gralow, J. R., Cardoso, F.,

Siesling, S., and Soerjomataram, I. (2022). Current

and future burden of breast cancer: Global statistics

for 2020 and 2040. The Breast, 66:15–23.

Ashreetha, B., V, D. G., Anandaram, H., A, N. B., Gupta,

N., and Verma, B. K. (2023). Iot wearable breast tem-

perature assessment system.

Elouerghi, A., Bellarbi, L., Khomsi, Z., Jbari, A., Errachid,

A., and Yaakoubi, N. (2022). A ﬂexible wearable ther-

mography system based on bioheat microsensors net-

work for early breast cancer detection: IoT technol-

ogy. Journal of Electrical and Computer Engineering,

2022:1–13.

Halim, A. A. A., Andrew, A. M., Yasin, M. N. M., Rahman,

M. A. A., Jusoh, M., Veeraperumal, V., Rahim, H. A.,

Illahi, U., Karim, M. K. A., and Scavino, E. (2021).

Existing and emerging breast cancer detection tech-

nologies and its challenges: A review. Applied Sci-

ences, 11(22):10753.

Hylton, N. (2005). Magnetic resonance imaging of the

breast: Opportunities to improve breast cancer man-

agement. Journal of Clinical Oncology, 23(8):1678–

1684.

Laila Fadhillah, U. D., Nur Aﬁkah, Z. A., Saﬁee, N. E. N.,

Asnida, A. W., and Raﬁq, M. (2018). Development

of a low-cost wearable breast cancer detection device.

pages 41–46.

Lee, A. H. (2005). Why is carcinoma of the breast more

frequent in the upper outer quadrant? a case series

based on needle core biopsy diagnoses. The Breast,

14(2):151–152.

Meijer, G. C., Wang, G., and Heidary, A. (2018). Smart

temperature sensors and temperature sensor systems.

In Smart Sensors and MEMs, pages 57–85. Elsevier.

Michaels, E., Worthington, R. O., and Rusiecki, J. (2023).

Breast cancer. Medical Clinics of North America,

107(2):271–284.

Ng, E., Acharya, U. R., Keith, L. G., and Lockwood, S.

(2007). Detection and differentiation of breast can-

cer using neural classiﬁers with ﬁrst warning thermal

sensors. Information Sciences, 177(20):4526–4538.

Ng, E. Y. K. (2001). Statistical analysis of healthy and ma-

lignant breast thermography. Journal of Medical En-

gineering & Technology, 25(6):253–263.

BIODEVICES 2024 - 17th International Conference on Biomedical Electronics and Devices

S, V. S., Royea, R., Buckman, K. J., Benardis, Fletcher,

R. L., Eyk, N., and Acharya, R. (2020). An introduc-

tion to the cyrcadia breast monitor: A wearable breast

health monitoring device. 197:105758.

Singh, D. and Singh, A. K. (2020). Role of image thermog-

raphy in early breast cancer detection- past, present

and future. 183:105074.

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

mataram, I., Jemal, A., and Bray, F. (2021). Global

cancer statistics 2020: Globocan estimates of inci-

dence and mortality worldwide for 36 cancers in 185

countries. CA: A Cancer Journal for Clinicians,

71(3):209–249.

Tao, Z., Shi, A., Lu, C., Song, T., Zhang, Z., and Zhao,

J. (2014). Breast cancer: Epidemiology and etiology.

Cell Biochemistry and Biophysics, 72(2):333–338.

Thigpen, D., Kappler, A., and Brem, R. (2018). The role of

ultrasound in screening dense breasts: A review of the

literature and practical solutions for implementation.

Diagnostics, 8(1):20.

Xiao, Y., Zhou, Q., and Chen, Z. (2015). Automated

breast volume scanning versus conventional ultra-

sound in breast cancer screening. Academic Radiol-

ogy, 22(3):387–399.

Yaffe, M. J. and Mainprize, J. G. (2011). Risk of radiation-

induced breast cancer from mammographic screening.

Radiology, 258(1):98–105.

Breast Cancer Detection Using Smart Wearable Devices with Thermal Sensors