This article has been accepted for publication in IEEE Access.

This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3420688

Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2023.1120000

IoMT Privacy Preservation: A Hash-Based DCIWT

Approach for Detecting Tampering in Medical
Data
CH. RUPA1 , (Senior Member, IEEE), SK. ARSHIYA SULTANA1 , R. PAVANA MALLESWARI1 , CH.
DEDEEPYA1 , THIPPA REDDY GADEKALLU2,3,4 , (Senior Member, IEEE) HYOUNG-KYU SONG5 ,
and Md. Jalil Piran6 , (Senior Member, IEEE)
1
VR Siddhartha Engineering College, Siddhartha Academy of Higher Education (Deemed to be) Vijayawada, Andhra Pradesh, India (e-mail:
[email protected], [email protected], [email protected], [email protected])
2
The college of Mathematics and Computer Science, Zhejiang, A&F University, Hangzhou 311300, China
3
Division of Research and Development, Lovely Professional University, Phagwara, India
4
Center of Research Impact and Outcome, Chitkara University, Punjab, India. (email: [email protected])
5
Department of Information and Communication Engineering and Convergence Engineering for Intelligent Drone, Sejong University, Seoul 05006, Republic of
Korea (email: [email protected])
6
Department of Computer Science and Engineering, Sejong University, 05006, Seoul, South Korea (e-mail: [email protected])
Corresponding author: H. K. Song ([email protected]) and Md. Jalil Piran ([email protected]).
This work was supported by the Institute of Information & Communications Technology Planning & Evaluation (IITP) under the
metaverse support program to nurture the best talents (IITP-2024-RS-2023-00254529) grant funded by the Korea government (MSIT) and
the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education
(2020R1A6A1A03038540)

ABSTRACT Smart Healthcare has brought advantages to both healthcare professionals and individuals
living in remote areas. However, deliberately or inadvertently altering medical records can result in an
inaccurate diagnosis. As a result of telemedicine, patients can easily communicate with their healthcare
providers, and the Internet of Medical Things (IoMT) devices enable remote consultations, allowing
for timely diagnosis and treatment. Medical data can be protected by watermarking, a combination of
steganography, and fingerprinting. This paper proposes a hash-based approach using the Discrete Cosine
Integer Wavelet Transform (DCIWT) to provide secure watermarking for confidential medical data. To
preserve critical Region of Interest (RoI) integrity, we use SHA-256 cryptographic hash algorithm. The
hashed Region of Interest (hRoI) is embedded in the Region of Non-Interest (RoNI) by DCIWT. To enhance
the segmentation process, we employ adaptive thresholding, which dynamically adjusts the threshold based
on local image characteristics, ensuring accurate segmentation of RoI and RoNI across varying image
qualities. The hidden data is significantly affected if the watermarked image is altered. As a result of
comparing the original RoI with the extracted RoI, we determine if an image has been altered. Tampered
regions are be recovered by extracting the RoI and using it for diagnosis. The proposed digital watermarking
technique maintains RoI integrity using message digests and histograms. Through extensive experiment
results, we demonstrate that the perceptual properties of the images are intact in terms of Peak Signal-to-
Noise Ratio (PSNR) and Structural Similarity Index (SSIM).

INDEX TERMS Embedding, Extraction, IoMT, Message Digest, Tamper detection, Telemedicine, Water-
marking

I. INTRODUCTION a physician [1]. Even though telemedicine has developed

quickly in recent years, some issues still exist. Among these
T HE Internet of Medical Things (IoMT) has greatly
benefited telemedicine among other areas. The use of
telecommunications technology allows patients who cannot
concerns is the possibility that medical information could
be utilised against patients or by third parties for their own
benefit.
reach a medical facility to receive health care services. A
telemedicine consult enables diagnostic images, videos, and Discrete Cosine Integer Wavelet Transform (DCIWT)
still images to be sent electronically for consultation by hash-based method is suggested to enable safe watermarking

VOLUME 11, 2023 1

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3420688

for private medical data. The suggested digital watermarking II. RELATED WORK
method preserves ROI integrity using message digests and This section provides a brief overview of recent developments
histograms. in digital watermarking. By assessing prior studies and ad-
In IoMT, privacy is crucial to protect patient information dressing their inadequacies, DCIWT preserves the privacy of
from leakage or cyberattacks, especially in telemedicine, an sensitive medical data.
application of IoMT. The risk of patient’s life may be in- Ravichandran et al. in [7] proposed a medical image wa-
creased if an attack occurs during the transfer of medical data termarking method using Integer Wavelet Transform (IWT),
over the Internet. Therefore, tampering with or altering medi- chaotic map, and SHA-256 for tamper recognition and re-
cal data can lead to misdiagnosis, incorrect treatment, identity trieval. RoNI IWT coefficients integrate a patient’s medical
theft, device malfunctions, privacy violations, or malicious history and their hospital logo for authentication and identifi-
exploitation. cation. Using SHA-256, this approach successfully retrieves
An IoMT is a network of connected medical devices and the altered blocks and identifies the tampered blocks with
sensors that collect and transmit health data. Approximately 100% accuracy. However, the drawback of this approach is
30% of all medical images are altered in some way, according that the RoNI pixels are destroyed during watermark extrac-
to the World Health Organization (WHO). Worldwide, 50 tion.
million medical records are manipulated or falsified each Hui et al. in [8] developed a reversible medical image wa-
year [2], [3]. In medical examinations, watermarking can be termarking algorithm for integrity authentication and privacy
used to conceal the contents. You can imperceptibly embed security. Based on the algorithm, which separates images into
a message within another message by using watermarking. regions of interest and non-interest, three types of watermarks
A patient’s personal information can be embedded in diag- are created. Medical data transmission between hospitals
nostic images to watermark medical data 4, 5, 6. Patients’ poses privacy risks. Active contour models are used to sepa-
privacy and anonymity can be protected while medical data rate biomedical images. Watermarks are created using a local
is securely transmitted. difference method using a hash function and encryption. The
By superimposing a watermark on an image, a watermark disadvantage is that it does not have an optimal computational
is applied digitally to protect it. Images are protected by efficiency.
watermarks, which serve as proof of authenticity. Tampering Solihah et al. in [9] proposed a self-embedding watermark-
includes adding an image, cropping an image, and removing ing system for authenticating medical photos and locating
a watermark. Under these specific scenarios, 99.5% of tam- tampered areas using chaotic encryption and Deoxyribonu-
pered images can be detected. cleic Acid (DNA) encoding. In order to identify and locate
tampered areas, two watermarks are created the authentica-
A. OUR CONTRIBUTIONS tion watermark and the localization watermark. A patient’s
The main contributions of this paper can be summarized as medical reports are divided into four blocks using chaotic
follows. encryption. The limitation of the proposed model is that it
cannot locate tampered areas accurately.
• Using Digital Imaging and Communications in
Rasha et al. in [10], developed a new tampered area de-
Medicine (DICOM) images, we design and implement
tection method to protect medical images transmitted over
a digital watermarking method that allows secure access
network systems. In order to diagnose a patient, it is crucial
over insecure networks.
to use RoI data to determine authenticity. The main objective
• To secure sensitive medical data, we use the crypto-
is to determine the RoI’s Tamper Detection Logic (TDL)
graphic hash function SHA-256.
authentication bits. The residual coefficient of the slantlet
• A tamper detection method is used to identify the altered
transform embeds them in the RoNI part. Research has proven
regions.
effective in detecting and identifying tampering in the RoI.
• To overcome their limitations, extensive comparison
This technique only supports a limited number of image
analysis work is done with existing approaches.
formats, which is a disadvantage.
• Using metrics such as Peak-Signal-to-Noise Ratio
Amira et al. in [11] proposed a secure watermarking
(PSNR), Structural Similarity Index (SSIM), and Mean
method for verifying the accuracy and reliability of medical
Square Error (MSE), the proposed mechanism ensures
reports based on frequency domains. Using Slantlet Trans-
the visual quality of the image.
form (SLT) and Discrete Cosine Transform (DCT) trans-
forms, the approach retrieves RoI’s relevant features and fuses
B. PAPER ORGANIZATION them with Enhanced Perceptual Rendering (EPR) to create
The remaining parts of this paper are organized as follows. a watermark. Watermarks are encoded using a chaotic map,
The literature review is presented in Section II. The proposed while signatures are created using a private key. Consistency
methodology is reviewed in Section III. Section IV discusses and reliability of the hash code of the recovered watermark
the results and analysis of the proposed work with medical are verified. The only downside is that it only supports JPEG
images. The conclusion is drawn in Section V. files.
Rishi et al. in [12] proposed a multifaceted watermarking
2 VOLUME 11, 2023

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3420688

technique to protect medical photos from unauthorized use Using the DCIWT, the proposed hash-based approach con-
and unwanted tampering or attacks. In the proposed approach, tributes to detecting tampering in medical data by generating a
the host image’s recovery information is compressed using compact and robust image hash that can be used to reconstruct
the Lempel-Ziv-Welch (LZW) algorithm, and a robust water- an image map and compare it with the received image to
mark is embedded using a transformation domain embedding identify tampered regions. IWT and DCT coefficients are
procedure. The approach, however, fails to locate altered used to generate a unique key for each image block using the
areas correctly. SHA-256 cryptographic hash algorithm. The RoI is embed-
Chia et al. in [13], developed a reversible data hiding ded into the RoNI using this key, and since it’s a one-way hash
method using Electromagnetic Navigation Bronchoscopy function, every image has a unique hash value, which ensures
with Imaging (ENMI) technology in which the frequency the watermark’s security and integrity.
of Electronic Patient Information (EPI) and Electronic Med- The process of comparing and determining whether the
ical Records (EMR) are transmitted over the Internet and image has been tampered with is as follows:
exchanged between authorized medical organizations. The • The original RoI is extracted from the image using the
expansion of each 2 × 2 block in the original image to a 3 × 3 key generated by the SHA-256 algorithm.
block is provided as a reversible data hiding strategy using • The extracted RoI is compared with the received RoI
improved neighborhood mean interpolation technology. The using a hash function that calculates a hash value for
detection performance is significantly higher. The primary each image block.
drawback of this approach is the low security for hidden data. • If the hash values of the original and received RoI blocks
The limitations of existing works include the destruction of are different, then the block is marked as tampered.
RoNI pixels during watermark extraction, suboptimal com- • The tampered blocks are then localized and recovered
putational efficiency, inability to locate tampered areas accu- using the compressed copy of the original RoI that is
rately, limited support for image formats, and low security embedded in the RoNI.
for hidden data. Our proposed hash-based approach using the
As the method uses a robust and secure hash function, it
Discrete Cosine Integer Wavelet Transform (DCIWT) over-
can detect even very small changes in the image content,
comes these issues by preserving RoI integrity with the SHA-
providing reliable evidence of tampering.
256 cryptographic hash algorithm and ensuring accurate seg-
A watermarked image can be authenticated by performing
mentation through adaptive thresholding. This method pro-
an Inverse Integer Wavelet Transform (IIWT), extracting the
vides enhanced security and supports accurate tampering de-
hash from the embedded RoNI, and comparing it with the
tection and recovery, maintaining the perceptual quality of the
original hash. If the extracted hash value matches the original
medical images. The proposed DCIWT technique performs
hash value, the image is considered authentic. A mismatch
better in tamper detection using hashing, can be applied to
indicates that the image has been altered, so the medical
various types of images, and produces better computational
data process should be discarded [21], [22]. This method
results. As shown in the Table 1, there are a number of existing
provides an efficient solution for watermarking medical im-
works.
ages using DCIWT and enables tamper detection using a
hashing approach, as well as distinguishing critical and non-
III. PROPOSED METHODOLOGY
critical regions within the medical data as RoI and RoNI,
In this paper, an architecture is proposed to address the is-
which are not addressed in existing works [23]–[25]. Hash-
sue of medical image tampering. Images can be captured
based DCIWT can help to address the challenges of tamper
and transmitted using Implantable Programmable Medica-
detection and privacy preservation in IoMT. A proposed tech-
tion Technology (IPMT) devices while a digital watermark
nique supported a variety of MRI image formats, including
ensures the images are correctly matched to the patient and
jpeg, png, jpg, and gif. Kaggle’s brain MRI dataset contains
protected from unauthorized access. In IoMT, a DCIWT wa-
MRIs of the brain collected from IoMT devices. The proposed
termark can be embedded in medical images or other data. It
system methodology is shown in Figure 1.
is possible to detect any unauthorized changes to the data by
To solve the identified problem in telemedicine diagnosis,
using the watermark. In addition to providing a secure method
the proposed method consists of five modules.
of authentication, hash-based DCIWT can also ensure that
data is only accessible to authorized individuals. The method 1) Segmentation of critical and non-critical regions,
distinguishes between critical and non-critical regions in the 2) Integration of Message Digest in RoI
images by using a segmentation approach. Cryptographic 3) Embedding Hashed RoI in RoNI
hash functions like SHA-256 are commonly used for their 4) Extraction of Embedded Data from RoNI
strong collision resistance [14]–[20]. UTF-8 is used to encode 5) Tamper Detection
the RoI message digest through DCT. By using different keys,
the hashing process adds a layer of security to the data. In A. SEGMENTATION OF CRITICAL AND NON-CRITICAL
order to prevent unauthorized access to the information, this REGIONS
method is used. In the non-critical region, IWT embeds the Segmenting an image into a Region of Interest (RoI) and a
hashed RoI (hRoI). Region of Non-Interest (RoNI) is crucial in medical image
VOLUME 11, 2023 3

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3420688

TABLE 1. Comparison Table for literature review

Research Proposed Approach Application Integrity check Analysis Parameters

Ravichandran et al. [9] IWT Medical data No SHA-256(Hashing)
Hui et al. [10] Active Contour Model Medical data Yes Encryption and Hashing fucntion
Solihah et al. [11] Chaotic encryption Patients medical reports Yes Authentication and Encryption
Rasha et al. [12] SLT Tampered area detection in medical images Yes Hashing
Amira [13] SLT,DCT,DWT verifying the medical reports Yes Not mentioned
Rishi et al. [14] Lempel-Ziv-Welch (LZW) algorithm Protecting medical photos from unauthorized use No Not applicable
Chia et al. [15] ENMI Electronic medical records No Not applicable

Algorithm 1: Segmentation Algorithm

IoMT

(<>)- embedding
(</>)-extraction
Input: Original medical image (IM ) from IoMT
Output: RoI (IRoI ) and RoNI (IRoNI )
RoI
Calculate hash
value(h)
using DCT
Hashed RoI
(hRoI)
1. Read the input medical image (IM ).
Segmentation of critical
and non-critical regions
2. Function GenerateRoI(IM ):
Original Image RoNI

using IWT
gray = rgbToGray(IM )
extracted RoI extracted
(RoI')
using IDCT
hashed RoI
(h'RoI)
using IIWT binaryimage = AdaptiveThreshold(gray); RoI =
h
Watermarked Image morphop (binaryimage )
h' Compare message digests h==h'

no
yes return RoI
extracted message digest (h')

Tampered Not tampered

3. Function GenerateRoNI(IM ,RoI ):

Restore
extracted RoI
RoNI = IM .copy()
RoNI[RoI.squeeze() != 0] = 0
FIGURE 1. Proposed Methodology return RoNI
4. Function AdaptiveThreshold(gray):
mean= gray.AdaptiveThreshold(gray, 255,
analysis. A medical image contains the crucial information gray.ADAPTIVE_THREAH_MEAN_C,
needed for treatment or diagnosis, while a RoNI contains less gray.THRESH_BINARY,11,2)
critical information. Thresholding is a simple and common return mean
method of segmenting RoI and RoNI. Based on the inten-
5. Function Main:
sity levels in the pixels of the image, setting a threshold,
RoI = GenerateRoI(IM )
converting the image to a binary image, and labeling the
RoNI = GenerateRoNI(IM , RoI)
RoI and RoNI accordingly [26]–[28]. Further refinement can
print(RoI, RoNI)
be achieved by post-processing steps such as morphological
return 0
operations or region-based methods. Algorithm 1 shows the
stepwise process of segmentation using adaptive threshold-
ing and morphological operations. The adaptive thresholding
method dynamically adjusts the threshold based on local im- value and incorporate it into the RoI. The DCT coefficient
age characteristics, making it more robust for varying image F(u, v) at frequency coordinates (u, v) is calculated as [22]:
qualities and conditions.
M −1 N −1
2 X X
F(u, v) = √ α(u)α(v) f (x, y)δ, (1)
B. INTEGRATION OF MESSAGE DIGEST INTO ROI MN x=0 y=0
Following the segmentation of the ROI, a hash value is pro-    
duced. A hash value, a distinct numerical number, is used (2x + 1)uπ (2x + 1)vπ
δ = cos cos , (2)
to symbolize a return on investment. The hash value was 2M 2N
produced using SHA-256, which guarantees that even slight 
modifications to the ROI produce distinct message digests.  √1 , for u = 0
This approach could be used to safely and impenetrably verify α(u) = qM1 (3)
M , for u > 0

the ROI integrity. The ROI’s message digest is encoded using
UTF-8 in the critical region. Here, the digital signal is repre- 
 √1 , for v = 0
sented by the hash value, and the host signal is represented
α(v) = qN1 (4)
by the ROI. By extracting the embedded hash value and
N , for v > 0

comparing it to the hash value that is now embedded in the
ROI, the integrity of the ROI might be confirmed at any time. where u and v− represent the frequency coordinates ranging
To integrate the encoded hash value into the RoI, DCT should and x and y represent the spatial coordinates within the block
be used. SHA-256 is used in algorithm 2 to generate the hash ranging.
4 VOLUME 11, 2023

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3420688

Algorithm 2: Integration of Hash into RoI Algorithm Algorithm 4: Extraction Algorithm

Input: Segmented RoI(RoI ) Input: Watermarked image (IW )
Output: Hashed RoI(hRoI ) Output: RoI(RoI ) and hash
1. Function GenerateHash(RoI ): 1. Function ExtractRoI(IW ):
hash = SHA256(RoI ) hRoI = IIWT(IW )
return hash he , RoI = IDCT(hRoI)
2. Function GenerateHashedRoI(hash, RoI ): hash = decode(he )
he = hash.encode(′ utf − 8′ ) return RoI, hash
hRoI = DCT(RoI, he ) 2. Function Main:
return hRoI ExtractRoI(IW )
3. Function Main: print ExtractRoI
GenerateHash(RoI ) return 0
GenerateHashedRoI(hash, RoI )
print GenerateHashedRoI
return 0
D. EXTRACTION OF EMBEDDED HROI FROM RONI
The embedded hRoI from the non-region of interest must
be recovered in order to maintain the original data. Con-
Algorithm 3: Embedding Algorithm
sequently, the embedding process is reversed, which in-
Input: (RoNI ) and Hashed RoI (hRoI ) volves applying the inverse of the Integer Wavelet Transform
Output: Watermarked Image (IW ) (IIWT). Integer Wavelet Transform in reverse is also known
1. Function as IIWT. Its main objective is to recreate the original signal
GenerateWatermarkedImage(hRoI ,RoNI ): from the wavelet coefficients obtained by the IWT. The IIWT
hRoIarray = np.array(hRoI)
can be represented mathematically as follows.
coefficients = filter(hRoIarray , scalingfilter ) X
IW = IWT(coefficients,RoNI) x[n] = h[k].w[n − k], (6)
return IW
where x[n] represents the restored signal, h[k] represents the
2. Function Main:
coefficients of the scaling filter and w[n − k] represents the
GenerateWatermarkedImage(hRoI ,RoNI )
wavelet coefficients. IWT produces a multiscale representa-
print GenerateWatermarkedImage
tion of data by transforming the original signal into wavelet
return 0
coefficients.
In the IIWT, scaling filters increase the sampling rate of
wavelet coefficients. The scaling filter is convolved with the
C. EMBEDDING HASHED ROI (HROI) INTO RONI increased wavelet coefficients to restore the signal. Wavelet
The process of integrating the RoI into the RoNI is enhanced coefficients can be used to reconstruct the original signal after
through the incorporation of the IhRoI . As a result, the SHA- this convolution process. To recover the original signal accu-
256 hash of the RoI is embedded into the original RoI, making rately, IIWT uses the same scaling filter as Integer Wavelet
it more secure and tamper-proof. By integrating the hRoI into Transforms. IWTs and IIWTs are affected by the choice of
the RoNI, it is widely used in image analysis to increase accu- scaling filter. Algorithm 4 shows how the hRoI is extracted
racy. Using this integration technique, each region’s features from the watermarked image.
are specified and the data is merged to form a comprehensive
E. TAMPER DETECTION
picture. Integer Wavelet Transform (IWT) is used to embed
hRoI into RoNI. The IWT is a mathematical tool used to Discrete Cosine Integer Wavelet Transform (DCIWT) tech-
compress images and retrieve information, and its inverse is nology can detect tampering with medical images that have
used to restore the originals. been watermarked. Detecting tampering involves extracting
The IWT can be expressed mathematically as follows. the hash value from the watermarked image and comparing it
X to the original. A match between the original hash value and
w[n] = x[k].h[n − k], (5) the extracted hash value indicates that the hashed image has
been altered. The IWT technique provides a reliable method
where w[n] represents the wavelet coefficients, x[k] represents for detecting tampering on medical images. The algorithm for
the original signal, and h[n − k] represents the scaling filter detecting tampering is shown in algorithm 5.
coefficients. At each decomposition level, the scaling filter
h[n-k] filters the original signal x[k], producing the wavelet IV. SIMULATION RESULTS AND DISCUSSIONS
coefficients w[n]. A hashed region of interest (hRoI) is em- This section contains a detailed description of the DCIWT
bedded into a RoNI using algorithm 3. mechanism and its results. Medical images can be secured
VOLUME 11, 2023 5

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3420688

Algorithm 5: Tamper Detection Algorithm (c),(g),(k),(o) displays the extracted RoI and its message
Input: Tampered watermarked Image (IwT ) digest from the watermarked image. Figure. 3 (d),(h),(l),(p)
Output: Tamper Detection exhibit the hash value and RoI of the tampered image. From
1. Function TamperDetection(IwT ): Figure. 3(c) and Figure. 3(d), it is evident that the message
RoIwT = GenerateRoI(IwT ) digests of the extracted hash and tampered hash are not equal.
h = GenerateHash( RoIwT ) By extracting the RoI embedded in the image, the tampered
h’ = ExtractRoI(IwT ) region can be located and recovered. Table 2 compares the
if h==h’ then existing techniques with the proposed approach. To evaluate
The image is tamper-free the proposed approach, Peak Signal-to-Noise Ratio (PSNR),
else Mean Square Error (MSE), and Structural Similarity Index
Image is tampered. Measure (SSIM) are used. The values of their considered
return 0 factors are summarized in Table 3 for MRI medical images.
The quality of an image generally improves with an increase
2. Function Main: in PSNR.Therefore, the proposed model provides secure
ExtractRoI(IwT ) medical image transmission without compromising image
print ExtractRoI quality. PSNR, MSE, and SSIM metrics can be calculated to
return 0 analyze the quality of specific medical images. The results
presented in Table 2 are the averaged results of four medical
images, ensuring a comprehensive evaluation of the proposed
method’s performance across different scenarios.
using hash-based approaches. This study focuses on DICOM  
images. According to the experimental results, the proposed m−1 n−1
1 X X
mechanism provides a high level of security and preserves im- MSE = (O(i, j) − D(i, j))2  , (7)
mn i=0 j=0
age quality. Furthermore, the proposed mechanism is resistant
(L − 1)2
 
to attacks and ensures the integrity of medical images during
PSNR =10log10 , (8)
transmission and storage because of the hash-based approach. MSE
Reversible watermarking ensures the integrity of the RoI by 
(2 ∗ µx ∗ µy + C1) ∗ (2 ∗ σxy + C2)

embedding a hash value of the RoI into the RoNI. In this SSIM = , (9)
(µ2x + µ2y + C1) ∗ (σx2 + σy2 + C2)
case, the hash value is generated by a secure hash function
that detects any changes in the content of the RoI. By using where O(i, j) represents the Original Image and D(i, j)
message digests and histograms, the technique also provides represents the watermarked image. m,n are the numbers of
two-level verification: Comparing the message digest with the rows and columns in the image. MSE displays the cumulative
extracted digest can verify the authenticity and ownership of squared error between the original and watermarked image,
the image. while PSNR displays the peak error. The smaller the MSE,
Histograms are graphical representations of digital images. the smaller the error.
This is a plot of the number of pixels for each tonal value. By using a reversible watermarking scheme, the proposed
An image’s gray level frequency represents the relative fre- approach preserves the perceptual quality of medical images
quency of occurrence of various gray levels. By comparing while embedding the watermark. The higher the watermark
the original histogram to the new one, the histogram can be payload, the lower the image quality and the higher the ro-
used to verify the quality and robustness of the image. These bustness of the watermark.
verification mechanisms provide various levels of protection The Structural Similarity Index measures how structurally
and detection against various types of attacks and distortions, similar two images are. The original and distorted images
contributing to the overall security and reliability of the sys- must be available for full-reference metrics. SSIM returns a
tem. Figure. 2 (a),(f),(k),(p) shows the original MRI Brain value between -1 and 1, with 1 indicating the images are iden-
images that are taken as input. Fig. 2 (b),(g),(l),(q) displays tical, and 0 indicating they are completely different. In (4) x,y
the segmented RoI from medical images which are treated as are the two images being compared. µx and µy are the mean
a watermark. Figure. 2 (c),(h),(m),(r) shows the watermarked pixel values of x and y. σx and σy are the standard deviations
images in which RoI concatenated with hash was embedded. of the pixel values of x and y, respectively. C1 and C2 are
Figure. 2 (d),(i),(n),(s) and Figure. 2 (e),(j),(o),(t) depict the constants to stabilize the division in case of small variances.
original and watermarked images’ histograms. The histogram Figure. 4 shows the Test cases performance analysis of Table
shows the difference in pixel values between the original MRI 3. Figure. 5 represents the comparison analysis with existing
image and the watermarked image. There is little difference systems.
between the original and watermarked histograms. The specific evaluation metrics or criteria used to assess
Figure. 3 (a),(c),(i),(m) shows the watermarked image in the performance of the proposed approach are:
which RoI concatenated with hash embedded into it. Fig- • Peak Signal-to-Noise Ratio (PSNR) and Structural Sim-
ure. 3 (b),(f),(j),(n) depicts the tampered image. Figure. 3 ilarity Index Measure (SSIM) measure the perceptual
6 VOLUME 11, 2023

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3420688

(a) (b) (c) (d) (e)

(f) (g) (h) (i) (j)

(k) (l) (m) (n) (o)

(p) (q) (r) (s) (t)

FIGURE 2. (a),(f),(k),(p) Original Image (b),(g),(l),(q) RoI (c),(h),(m),(r) Watermarked Image (d),(i),(n),(s) Original Image Histogram (e),(j),(o),(t) Watermarked
image histogram

quality and similarity of the watermarked and original detecting tampering in medical data are:
images. • Using a dataset of 100 medical images of different
• Normalized Correlation (NC), which measures the ro- modalities, such as CT, MRI, X-ray, etc.
bustness and security of the watermark against various • Embedding a watermark of 16 kb into each image using
attacks and distortions. the proposed approach.
• Tampering Detection Rate (TDR) and Tampering Lo- • Applying various attacks and distortions to the water-
calization Rate (TLR), which measure the accuracy and marked images, such as cropping, rotation, scaling, com-
efficiency of the tampering detection and localization pression, noise, filtering, etc.
algorithm. • Extracting the watermark and comparing it with the
• The experimental setup or methodology used to validate original watermark using NC.
the effectiveness of the hash-based DCIWT approach for • Detecting and localizing the tampered regions using the

VOLUME 11, 2023 7

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3420688

(a) (b) (c) (d)

(e) (f) (g) (h)

(i) (j) (k) (l)

(m) (n) (o) (p)

FIGURE 3. (a),(c),(i),(m) - Watermarked Image (b),(f),(j),(n) -Tampered Image (c),(g),(k),(o) - Extracted hash (d),(h),(l),(p)- Tampered hash

hash-based DCIWT algorithm. Enhancing watermark robustness means making it stronger

• Comparing and evaluating the watermarked image and against different changes. We use methods like spreading
the original image using PSNR and SSIM. the watermark signal, adjusting it based on how humans
• Evaluating the tampering detection and localization per- see, adding extra bits for error recovery, and embedding it
formance using TDR and TLR. in multiple ways to ensure it stays intact despite various
transformations.

• Spread Spectrum Embedding: Spread spectrum tech-

niques spread the watermark signal across the image,
making it resilient to common transformations. By em-
bedding the watermark in multiple frequency bands, it
can withstand distortions such as resizing or compres-
8 VOLUME 11, 2023

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3420688

TABLE 2. COMPARISON ANALYSIS WITH THE EXISTED TECHNIQUES

Factors IWT SLT DCT Proposed

[9]–[11] [15]–[17] [20]–[22] Methodol-
ogy(DCIWT)
PSNR 57.84 60.41 N/A 56.752
MSE 0.694 0.7225 N/A 0.9545
SSIM 0.794 0.804 N/A 0.9825
Accuracy 80% 73% 86% 94%
Embedded 0.03 0.56 0.08 0.0023
time
Extracted 0.28 0.98 0.35 0.067
time
Integrity N/A Yes Yes Yes
Check
Encryption N/A N/A Yes N/A
Time O(n2 ) O(n3 ) O(n) O(log n) FIGURE 5. Comparing analysis with existing systems
Com-
plexity

tion level. By focusing watermark energy in perceptually

important regions, they enhance robustness while mini-
mizing visual degradation.
• Multi-Domain Watermarking: This approach embeds
the watermark in multiple domains (e.g., spatial, fre-
quency, or wavelet domains) simultaneously. By diver-
sifying the embedding, it improves resilience against a
wide range of transformations, ensuring the watermark’s
integrity across various processing operations.

V. CONCLUSION
Digital watermarking of medical images from IoMT ensures
the integrity and authenticity of medical images while de-
FIGURE 4. Test cases performance analysis tecting any tampering. In conjunction with IWT, DCT digital
watermarking provides a subtle and non-intrusive way to
embed the hash value into the RoI, making it ideal for medical
sion. imaging. Through the integration of hRoI, RoNI provides a
• Quantization Index Modulation (QIM): QIM embeds the more robust and reliable representation of critical diagnostic
watermark by modulating the quantization levels of the information. Scaling filters are used to restore the original
host signal. It offers robustness against transformations signal from the magnified coefficients. With additional layers
by exploiting the non-uniformity of human perception in of security, image data can be validated and ensured to be
different frequency bands. accurate, improving the precision of analysis and decision-
• Error Correction Coding: Error correction codes add making. By embedding a hashed message digest of the critical
redundancy to the watermark, allowing for the recovery region of the medical image into the non-critical region,
of damaged or distorted parts during watermark extrac- the proposed approach addresses the challenge of preserving
tion. Techniques like Reed-Solomon codes can correct privacy in IoMT. As a result, the image is protected from
errors caused by transformations such as noise addition unauthorized access and tampering and properly matched to
or geometric distortions. the patient. Moreover, a segmentation approach is used to
• Perceptual Models: These models simulate human vi- distinguish critical and non-critical regions in the images, and
sual perception to adjust the strength of the watermark a hash-based authentication scheme is used to verify their in-
according to the image content and the expected distor- tegrity. In the telemedicine field associated with the internet,
these mechanisms provide an efficient and secure solution for
tamper detection. To protect copyright or verify authenticity,
TABLE 3. Test Cases Performance Analysis the project can be extended to all types of multimedia and
IoMT-based medical images in the future.
Original Watermarked PSNR MSE SSIM
image image
ACKNOWLEDGMENT
Fig 2(a) Fig 2(c) 58.721 1.521 0.99
Fig 2(f) Fig 2(h) 57.362 1.390 0.98 This work was supported by the Institute of Informa-
Fig 2(k) Fig 2(m) 57.143 0.971 0.99 tion & Communications Technology Planning & Evalua-
Fig 2(p) Fig 2(r) 55.285 0.937 0.97 tion (IITP) under the metaverse support program to nurture
VOLUME 11, 2023 9

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3420688

the best talents (IITP-2023-RS-2023-00254529) grant funded [19] Y. Nagasree, C. Rupa, A. Akshitha, P. Srivastava, T.R. Gadekallu, "Preserv-
by the Korea government (MSIT), and the Basic Science ing Privacy of Classified Authentic Satellite Lane Imagery Using Proxy
Re-Encryption and UAV Technologies," Drones 2023, vol. 7, 2023.
Research Program through the National Research Founda- [20] M. Tanaka, H. Kiya, "Fake-image detection with Robust Hashing," in IEEE
tion of Korea (NRF) funded by the Ministry of Education 3rd Global Conference on Life Sciences and Technologies (LifeTech), pp.
(2020R1A6A1A03038540) 40–43, 2021.
[21] C. Rupa, M. Harshita, G. Srivastava, T. R. Gadekallu, P. K. R. Maddikunta,
"Securing Multimedia using a Deep Learning based Chaotic Logistic
REFERENCES Map," IEEE Journal of Biomedical and Health Informatics, 2022.
[22] Wenbo Wan, Jun Wang, Yunming Zhang, Jing Li, Hui Yu, Jiande Sun, "A
[1] P. Singh, K. J. Devi, H. K. Thakkar, and K. Kotecha, "Region-based hybrid comprehensive survey on robust image watermarking," Neurocomputing,
medical image watermarking scheme for robust and secured transmission vol. 488, pp. 226–247, 2022.
in IoMT," IEEE Access, vol. 10, pp. 8974-8993, 2022. [23] D. Fan, X. Zhang, W. Kang, H. Zhao, Y. Lv, "Video Watermarking Algo-
[2] Q. Li, Y. Fu, and Z. Zhang, "Medical images lossless recovery based rithm Based on NSCT, Pseudo 3D-DCT and NMF," Sensors, vol. 22, pp.
on POB number system and image compression," in Multimedia Tools 1–18, 2022.
Applications, pp. 11415-11440, 2022. [24] A. Anand, A. K. Singh, "Secret sharing based watermarking for copy-
[3] K. Yu, L. Tan, S. Mumtaz, S. Al-Rubaye, A. Al-Dulaimi, A. K. Bashir, and protection and ownership control of medical Image," in 12th International
F. A. Khan, "Securing critical infrastructures: Deep-learning-based threat Conference on Computing Communication and Networking Technologies
detection in IIoT," IEEE Communications Magazine, vol. 59, pp. 76–82, (ICCCNT), pp. 01-07, 2021.
2021. [25] Z. Liu, J. Li, Y. Ai et al., "A robust encryption watermarking algorithm for
[4] T. Sajjad Ali, Rashid Ali, "A novel medical image encryption scheme using medical images based on ridgelet-DCT and THM double chaos," J Cloud
TLTS and Henon chaotic map," IEEE Access, vol. 8, pp. 71974–71992, Comp 11, vol. 60, 2022.
2020. [26] D. B, C. Rupa, H. D, V. Y., "Privacy Protection of Digital Information
[5] N. S. Utami, L. Novamizanti, S. Saidah, I. N. Apraz Ramatryana, "Robust using Frequency Domain Watermarking Technique," in 2021 4th Interna-
color image watermarking technique in the spatial domain," in IEEE tional Conference on Recent Trends in Computer Science and Technology
International Conference on Industry 4.0, Artificial Intelligence, and Com- (ICRTCST), vol. 60, pp. 202-206, 2021.
munications Technology (IAICT), pp. 32–38, 2021. [27] P. Singh, K. J. Devi, H. K. Thakkar, K. Kotecha, "Region-Based Hybrid
[6] A. Shankar, A. Kannammal, "A Hybrid Of Watermark Scheme With En- Medical Image Watermarking Scheme for Robust and Secured Transmis-
cryption To Improve Security Of Medical Images," in Third International sion in IoMT," IEEE Access, vol. 10, pp. 8974-8993, 2022.
Conference on Intelligent Communication Technologies and Virtual Mobile [28] A. Alzahrani, N. A. Memon, "Blind and Robust Watermarking Scheme
Networks (ICICV), pp. 226–233, 2021. in Hybrid Domain for Copyright Protection of Medical Images," IEEE
[7] A. Anand, A. K. Singh, "Health Record Security Through Multiple Water- Access, vol. 9, pp. 113714-113734, 2021.
marking on Fused Medical Images," IEEE Transactions on Computational [29] M. Charfeddine, E. Mezghani, S. Masmoudi, C. B. Amar, H. Alhumyani,
Social Systems, vol. 9, pp. 1594–1603, 2022. "Audio Watermarking for Security and Non-Security Applications," IEEE
Access, vol. 10, pp. 12654-12677, 2022.
[8] F. Yan, H. Huang, X. Yu, "A Multiwatermarking Scheme for Verify-
[30] Li, W., Susilo, W., Xia, C., Huang, L., Guo, F.,... Wang, T. (2024). Secure
ing Medical Image Integrity and Authenticity in the Internet of Medical
Data Integrity Check Based on Verified Public Key Encryption with Equal-
Things," IEEE Transactions on Industrial Informatics, vol. 18, pp. 8885–
ity Test for Multi-Cloud Storage. IEEE Transactions on Dependable and
8894, 2022.
Secure Computing. doi: 10.1109/TDSC.2024.3375369.
[9] D. Ravichandran, P. Praveenkumar, S. Rajagopalan et al., "ROI-based
[31] D. B, C. Rupa, H. D, V. Y., "Privacy Preservation of Medical Health
medical image watermarking for accurate tamper detection, localization,
Records using Symmetric Block Cipher and Frequency Domain Water-
and recovery," Med Biol Eng Comput, vol. 59, pp. 1355–1372, 2021.
marking Techniques," in 2022 International Conference on Inventive Com-
[10] H. Shi, Y. Wang, Y. Li, "Region-based reversible medical image water- putation Technologies (ICICT), vol. 10, pp. 96-103, 2022.
marking algorithm for privacy protection and integrity authentication," [32] M. Magdy, N. I. Ghali, S. Ghoniemy, K. M. Hosny, "Multiple Zero-
Multimed Tools Appl, vol. 80, pp. 24631-24667, 2021. Watermarking of Medical Images for Internet of Medical Things," IEEE
[11] S. Gull, R. F. Mansour, N. O. Aljehane et al., "A self-embedding technique Access, vol. 10, pp. 38821-38831, 2022.
for tamper detection and localization of medical images for smart-health," [33] J. Li, J. Li, C. Wang, F. J. Verbeek, T. Schultz, H. Liu, "Outlier detection
Multimedia Tools and Applications, vol. 80, pp. 29939–29964, 2021. using iterative adaptive mini-minimum spanning tree generation with ap-
[12] Rasha Thabit, "TAMPER DETECTION AND LOCALIZATION plications on medical data," Frontiers in Physiology, vol. 14, 2023, doi:
SCHEME FOR COLOR MEDICAL IMAGES," Journal of Engineering 10.3389/fphys.2023.1233341.
Science and Technology, vol. 17, pp. 0760–0780, 2022. [34] Ch, R., Yadlapalli, V., Sk, S. S., Reddy, G. T.,Kautish, S. Robust stegano-
[13] Amira K. Jabbar, Ashwaq T. Hashim, Qussay F. Hassan, "Medical Image graphic framework for securing sensitive healthcare data of telemedicine
Authentication by Combining Hash Signature and Watermarking Based using convolutional neural network.CAAI Transactions on Intelligence
on Frequency Domains," in 2nd International Conference on Physics and Technology, https://doi.org/10.1049/cit2.12319.
Applied Sciences (ICPAS 2021), vol. 1963, 2021. [35] Jiang, H., Wang, M., Zhao, P., Xiao, Z., Dustdar, S. (2021). A Utility-
[14] Rupa Ch, Naga Vivek K, Gautam Srivastava, Reddy Gadekallu, Aware General Framework With Quantifiable Privacy Preservation for
ECDSA-based tamper detection in medical data using a Destination Prediction in LBSs. IEEE/ACM Trans. Netw., 29(5), 2228-
watermarking technique, International Journal of Cognitive 2241. doi: 10.1109/TNET.2021.3084251.
Computing in Engineering, Volume 5,2024,Pages 78-87,ISSN 2666- [36] N. V. R. G, M. Ch, R. Ch and A. B. N, "Hash Based Secure
3074,https://doi.org/10.1016/j.ijcce.2024.01.003. and Efficient Scaling Approach for Detecting the Diabetic Retinopa-
[15] R. Sinhal, S. Sharma, I. A. Ansari, "Multipurpose medical image water- thy Disorder," 2024 International Conference on Emerging Smart
marking for effective security solutions," Multimed Tools Appl, vol. 81, pp. Computing and Informatics (ESCI), Pune, India, 2024, pp. 1-6, doi:
14045–14063, 2022. 10.1109/ESCI59607.2024.10497243.
[16] Hia-Chen Lin, Chin-Chen Chang, Wei-Jiun Kao, Jui-Feng Chang, "Effi- [37] Li, M., Cui, H., Liu, C., Shan, C., Du, X.,... Guizani, M. (2023). A Four-
cient Electronic Patient Information Hiding Scheme With Tamper Detec- Dimensional Space-Based Data Multi-Embedding Mechanism for Net-
tion Function for Medical Images," IEEE Access, vol. 10, pp. 45474-45485, work Services. IEEE Transactions on Network and Service Management.
2022. doi: 10.1109/TNSM.2023.3339674.
[17] E. C. P, R. Ch, N. V. K, C. K, "Privacy Preservation of Plant Disease
Detection using Hashing based Convolution Neural Network," in 2022 3rd
International Conference on Computing, Analytics and Networks (ICAN),
pp. 1–6, 2022.
[18] S. Gadamsetty, R. Ch, A. Ch, C. Iwendi, T.R. Gadekallu, "Hash-Based
Deep Learning Approach for Remote Sensing Satellite Imagery Detec-
tion," Water 2022, vol. 14, 2022.

10 VOLUME 11, 2023

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3420688

CH. RUPA is currently working as a professor with THIPPA REDDY GADEKALLU is currenty work-
VRSEC, Vijayawada. She has published more than ing with the Division of Research and Develop-
100 articles in various journals and conferences. ment, Lovely Professional University, Phagwara,
Her main research interests include information India as well as with the Center of Research Impact
security, blockchain, machine learning, AI based and Outcome, Chitkara University, Punjab, India.
security approaches. She is a life member of CSI, He obtained his Bachelors in Computer Science
ISTE, IAENG, IEI, and IACSIT. She was listed in and Engineering from Nagarjuna University, India,
the top 2 percent of scientists by Stanford Univer- in the year 2003, Masters in Computer Science
sity and Elsevier. She also bagged many national and Engineering from Anna University, Chennai,
and international awards and recognitions, like Se- Tamil Nadu, India in the year 2011 and his Ph.D
nior member of IEEE, Young Women Engineer, Technomedia, etc by IEI, in Vellore Institute of Technology, Vellore, Tamil Nadu, India in the year
IETE, etc. 2017. He has more than 15 years of experience in teaching. He has more
than 200 international/national publications in reputed journals and confer-
ences. Currently, his areas of research include Machine Learning, Internet
of Things, Deep Neural Networks, Blockchain, Computer Vision. He is
an editor in several publishers like Springer, Hindawi, Plosone, Scientific
Reports (Nature), Wiley. He also acted as a guest editor in several reputed
publishers like IEEE, Elsevier, Springer, Hindawi, MDPI. He is recently
recognized as one among the top 2% scientists in the world as per the survey
conducted by Elsevier in the years 2021, 2022, 2023. He is also recognized
as a highl;y cited researcher, young scienist category, by Clarivate (web of
Science) for the period 2017-2022 in the year 2023.
SK. ARSHIYA SULTANA has received gold model
in her graduation under JNTUK. She is currently
pursuing her research work in VRSEC (A). She has
published her works in the SCI and IEEE indexed
Conferences. Her research areas are cryptography,
watermarking, and blockchain technology. She has
completed many certification courses and partici-
pated in workshops related to her research area.

R. PAVANA MALLESWARI is currently pursuing

her research work in VRSEC (A) under the Com-
puter Science and Engineering Course. She has
published her work in the SCI and IEEE-indexed
conferences. Her research areas are cryptography,
watermarking, and blockchain technology. She has
completed many certification courses and partici-
pated in workshops related to her research area.

HYOUNG-KYU SONG received the B.S., M.S.,

and Ph.D. degrees in electronic engineering from
Yonsei University, Seoul, South Korea, in 1990,
1992, and 1996, respectively. From 1996 to 2000,
CH. DEDEEPYA currently working as a Packaged he was a Managerial Engineer with the Korea Elec-
App Development Associate in ACCENTURE. tronics Technology Institute (KETI), South Ko-
She received her engineering degree from VR- rea. Since 2000, he has been a Professor with the
SEC (A) in computer science and engineering. She Department of Information and Communication
has published her research work in IEEE confer- Engineering and the Department of Convergence
ences and Scopus journals. Her research areas are Engineering for Intelligent Drone, Sejong Univer-
blockchain technology, watermarking, and infor- sity, Seoul. His current research interests include digital and data commu-
mation security. nications, information theory, and their applications, with an emphasis on
mobile communications.

VOLUME 11, 2023 11

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3420688

MD. JALIL PIRAN (S’10, M’16, SM’21) holds a

distinguished academic background and currently
serves as an Associate Professor at the Depart-
ment of Computer Science and Engineering in
Sejong University, Seoul, South Korea. He earned
his Ph.D. degree in Electronics and Information
Engineering from Kyung Hee University, South
Korea, in 2016, and subsequently worked as a Post-
Doctoral Fellow at the Networking Laboratory of
the same institution.
Prof. Piran has made significant contributions to the field of Artificial
Intelligence and Data Science through his extensive research publications
in esteemed international journals and conferences. His areas of expertise
encompass Machine Learning, Data Science, Big Data, Internet of Things
(IoT), and Cyber Security. In addition to his research endeavors, Prof. Jalil Pi-
ran actively engages with scholarly journals as an Editor, including the "IEEE
Transactions on Intelligent Transportation Systems," "Elsevier Journal of
Engineering Applications of Artificial Intelligence," "Elsevier Journal of
Physical Communication," and "Elsevier Journal of Computer Communica-
tion." He also serves as Secretary of the IEEE Consumer Technology Society
on Machine Learning, Deep Learning, and AI. Furthermore, he assumes the
role of Track Chair for Machine Learning, Deep Learning, and AI in the
CE (MDA) Track for the upcoming 2024 IEEE International Conference
on Consumer Electronics (ICCE). In 2022, he chaired the "5G and Beyond
Communications" Session at the prestigious IEEE International Conference
on Communications (ICC). Prof. Piran is esteemed as a Senior Member of
IEEE and represents South Korea as an Active Delegate to the Moving Picture
Experts Group (MPEG). His outstanding research contributions have been
recognized internationally, as evidenced by the prestigious "Scientist Medal
of the Year 2017" awarded by IAAM in Stockholm, Sweden. Moreover, he
received accolades from the Iranian Ministry of Science, Technology, and
Research as an "Outstanding Emerging Researcher" in 2017. His exceptional
Ph.D. dissertation was honored as the "Dissertation of the Year 2016" by
the Iranian Academic Center for Education, Culture, and Research in the
Engineering Group.

12 VOLUME 11, 2023

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4