2023 3rd International Conference on Intelligent Technologies (CONIT)

Karnataka, India. June 23-25, 2023

Securing Medical Images using Compression

techniques with encryption and Image Steganography
Pooranakala K. Dr.Vanita Jaitly
2023 3rd International Conference on Intelligent Technologies (CONIT) | 979-8-3503-3860-7/23/$31.00 ©2023 IEEE | DOI: 10.1109/CONIT59222.2023.10205855

PG Student, Associate Professor

Department of Computer Applications Department of Computer Applications
Hindustan Institute of Technology and Science Hindustan Institute of Technology and Science
Chennai, Tamil Nadu, India Chennai, India
[email protected] [email protected]

Abstract- Medical imaging is an essential component of the confidentiality, integrity, and authentication. Therefore,
healthcare sector, enabling the transfer of medical images safeguarding the confidentiality and integrity of medical
between widely separated locations to improve healthcare images during their transmission is crucial and must be given
business practices. However, the transmission of medical images the highest priority. This study proposes a novel method for
across public networks can result in several security concerns, securing medical images during their transmission over public
including those related to authentication, integrity, and networks. The proposed method comprises three steps:
confidentiality. Therefore, the major goal of this research is to compression, encryption, and embedding. In the first step, the
ensure the security and accuracy of medical images by image undergoes compression using the SVD technique,
maintaining their confidentiality and integrity. This proposed
resulting in a smaller image size and noise elimination. The
method aims to enhance the security of medical images during
their transmission across public networks. The method involves
second step implies encrypting the compressed image using the
three main steps: compression, encryption, and embedding To AES algorithm, providing robust encryption to maintain
achieve the goal of maintaining the integrity and confidentiality confidentiality. Finally, the encrypted image is securely
of medical images, this research utilizes a multi-step process. embedded within a cover image using the LSB steganographic
First, the medical image undergoes compression using the technique, creating a stego image. The SVD technique enables
Singular Value Decomposition (SVD) technique to reduce its size the compression and denoising of medical images, while the
and eliminate any noise. Following this, the compressed image is AES algorithm provides strong encryption to ensure
encrypted using the Advanced Encryption Standard (AES) confidentiality. To ensure the integrity of the medical image
algorithm to ensure confidentiality. Finally, the encrypted image during transmission, the encrypted image is seamlessly and
is securely enclosed within a cover image using the Least securely embedded within the cover image using the LSB
Significant Bit (LSB) steganographic technique to create a stego steganographic technique, rendering the changes imperceptible.
image. This technique ensures that the encrypted image is The proposed method offers a promising solution for securing
seamlessly incorporated within the cover image, without any medical images during their transfer over public networks. The
visible or perceptible changes. This method ensures that the experimental results confirm the efficacy of the proposed
medical image remains confidential and maintains its integrity method in enhancing the security of medical images while
throughout the transmission process. Furthermore, this study preserving their quality.
developed the aforementioned technique, and the experimental
outcomes were compared using MATLAB to hide medical images II. LITERATURE SURVEY
without encryption using LSB steganography as an additional
layer of security. This study analyzed multiple medical image While usually either cryptography or steganography is
formats, including DICOM, TIFF, BMP, and JPEG. The findings employed to ensure security, in this instance, both techniques
indicate that combining encryption and steganography were integrated to provide enhanced security [1]. To achieve
techniques provides improved results in terms of PSNR and MSE this, transform domain techniques such as DWT and DCT can
values, as contrasted to utilizing steganography alone. be employed, and the performance can be assessed based on
various parameters [2]. Compression and cryptographic
Keywords— Encryption, Steganography, LSB, SVD, Medical techniques are utilized to secure text files. Compression refers
images,Security to the process of reducing the number of bits necessary to
represent a particular set of data. It is helpful in saving more
I. INTRODUCTION data. To develop a digital data security application, the
Medical images, such as X-rays, MRI scans, and CT scans, combination of the DES (Data Encryption Standard) algorithm
contain sensitive and personal information that must be with the cryptographic method and the steganographic method
protected from unauthorized access. The proposed approach with Discrete Cosine Transform (DCT) can be used [3].
provides a secure and efficient solution for storing and
The quality of the stego-image can be concluded based on
transmitting medical images. Medical imaging is an integral
the experiment. In order to improve the security of data files
part of healthcare that enables healthcare professionals to share
before being sent over the public network, a data security
medical images between geographically separated locations.
application is provided by this research. Here four levels of
However, the transfer of medical images over public networks
security technique for veiling messages are proposed which
can pose various security risks, including issues related to
utilize two levels of cryptography with two levels of

979-8-3503-3860-7/23/$31.00 ©2023 IEEE 1

Authorized licensed use limited to: GITAM University. Downloaded on February 10,2025 at 05:44:40 UTC from IEEE Xplore. Restrictions apply.
steganography is used [4]. Vernam cipher is modified at each seems promising and could lead to better results compared to
level of cryptography with the automatically originating initial traditional fixed percentage methods for selecting singular
key from the random pixel of camouflage cover. Better values in SVD-based image compression.
camouflage is created to evade interloper attention and realize
better performance in terms of steganographic system III. PROPOSED METHODOLOGY
measurement.[5]. To enhance the system's robustness, we have The proposed method for securing medical images during
concentrated on improving the security and steganography transmission involves a three-step process. First, the medical
processes. Specifically, we have focused on using the LSB image is compressed using the SVD approach. Then, The AES
method and RGB for the steganography operation. In one algorithm is used to encrypt the compressed image. Finally, the
technique of image steganography, we have emphasized the encrypted image is embedded inside a cover image using the
MSB bits rather than the typical LSB bits [6]. The LSB steganographic technique, resulting in a stego image that
experimental findings indicate that the modified method is can be transferred via an uncertain channel. At the receiver
suitable for embedding a small quantity of data. Several of the end, the AES algorithm is used to decrypt the encrypted
methodologies employed in the study rely on the results of the version of the medical image, thereby obtaining the original
DES, which is a symmetric key algorithm in cryptography [7]. medical image. The use of cryptography provides
It can be concluded that the combination of these two confidentiality to the medical image, as only authorized parties
cryptographic techniques can transmit data through the with access to the decryption key can view the original image.
unsecured channel and provides confidentiality for information The use of steganography provides additional security by
hiding. One of the potential limitations of this proposed system hiding the encrypted medical image within a cover image. The
is that the input data size may increase, resulting in a larger utilization of steganography to embed the medical image
position array size.[8] SVD is capable of breaking down a within a cover image makes it challenging for unauthorized
matrix into three constituent matrices: a diagonal matrix of parties to identify the presence of the medical image.
singular values, a left orthogonal matrix, and a right orthogonal Nevertheless, it should be acknowledged that solely relying on
matrix. steganography for security purposes does not guarantee
In the field of digital image processing, By minimizing the confidentiality since individuals with access to the stego image
amount of singular values used to represent an image, SVD can and appropriate steganography tools can extract the hidden
be used to compress images. This is achieved by retaining only medical image. Overall, the proposed method seems to provide
the most significant singular values and discarding the rest. a good combination of cryptography and steganography to
The resulting compressed image will have a lower resolution achieve both integrity and confidentiality for medical images.
and quality but will require less storage space in memory. The retrieved medical image is depicted in Figure 1.

Experiments can be conducted to evaluate the performance

of different singular values for image compression. The
performance evaluation parameters can include metrics such as
image quality, compression ratio, and processing time. The
updated method for selecting the singular values that are
retained with the related singular vectors in SVD picture
compression is proposed in this research. [10] The goal of this
technique is to keep the image's edges, which represent a
highly important. visible component to human eyes. This is
accomplished by permitting a range of percentages, calculated Fig. 1. shows the combination of encryption and steganography
as the total of the singular values that are disregarded, as
opposed to a fixed percentage. The variance is used to gauge Top of Form
how edgy a block is, with the block with the highest variance 1) Cryptography AES algorithm:
receiving the lowest percentage within the given range. The The AES algorithm is an effective method for encrypting
research aims to improve image compression using SVD by and decrypting various forms of data, including images. The
proposing a new method for selecting the singular values that fundamental concept behind AES encryption involves taking a
are retained with the related singular vectors. The goal is to plaintext image and transforming it into ciphertext using a key.
preserve the image's edges, which are important visible Subsequently, the same key is utilized to decrypt the ciphertext
components to the human eye [11]. The proposed method uses and recover the original plaintext image.
a variable percentage range rather than a fixed percentage to
retain the singular values. The range is calculated based on the Here's a high-level overview of the steps involved in AES
total of the singular values that are discarded. The variance of encryption for images:
the image blocks is used to measure the edge content of each 1. Convert the plaintext image into a stream of bits.
block. Blocks with higher variance (i.e., more edges) are given
a lower percentage within the variable range, ensuring that the 2. Divide the stream of bits into 128-bit blocks, the
edges are better preserved in the compressed image. block size used by AES.
3. Initialize a 128-bit key, which will be used to encrypt
Overall, this proposed method aims to improve image
compression quality by better preserving the edges in the and decrypt the image.
image, which are important visual features. The approach 4. Use the key to encrypt each block of bits using the

2
Authorized licensed use limited to: GITAM University. Downloaded on February 10,2025 at 05:44:40 UTC from IEEE Xplore. Restrictions apply.
 AES encryption algorithm. A_ approx. = U * S * V^T
5. Concatenate the encrypted blocks to form the 3) Image Compression:
ciphertext. The low-rank approximation of the image matrix can be
AES is a widely used and well-regarded encryption further compressed by removing redundant information. This
standard, and it is very secure when used correctly. However, can be achieved by quantizing the image or by using lossless or
like all encryption algorithms, it can be vulnerable to attack if lossy compression techniques. The compression ratio achieved
the key is not properly secured. To ensure the security of your by the SVD algorithm is directly related to the number of
encrypted image, it is important to choose a strong key, use singular values retained in the low-rank approximation. When
proper key management procedures, and store the key securely. more singular values are retained, the compression ratio is
higher but the image quality is lower. Conversely, when fewer
singular values are retained, the compression ratio is lower but
the image quality is higher. Therefore, the choice of the
number of singular values to retain is a trade-off between the
compression ratio and the image quality, and it depends on the
specific application requirements.
a) Image Reconstruction:
The reconstructed image obtained from the low-resolution
and denoised image using the SVD technique is suitable for a
variety of image-processing applications, including data
analysis and compression.

Fig. 2. AES algorithm encryption

B. SVD Compression:
The SVD algorithm is a powerful mathematical tool for
image processing. It is used to analyse and manipulate the
structure of a matrix, which represents an image. The algorithm
provides a low-rank approximation of the original matrix,
which reduces the resolution of the image and removes noise.
The following is a detailed explanation of the working of the
SVD algorithm and what results in it provides.
1) Matrix Decomposition:
A mathematical method called the SVD algorithm is used to
divide matrix A into the three matrices U, S, and V. Although
matrix S is a diagonal matrix that contains the singular values
of A in descending order, matrix U is an orthogonal matrix
that contains the left singular vectors of A. The right singular
vectors of A are contained in the orthogonal matrix V. These
matrices are obtained by performing an eigen decomposition of
the matrices A^T A and A A^T. In the context of digital image
processing, By decreasing the number of singular values
needed to represent the image, the SVD technique can be used
to compress images.
2) Low-Rank Approximation:
To obtain a low-rank approximation of the original matrix
A using the SVD algorithm, we need to retain only the Fig. 3. SVD Compression
corresponding singular vectors and largest singular values.
This low-rank approximation can effectively reduce the The SVD algorithm provides a powerful tool for image
resolution of the image and remove noise. The matrices U, S, processing by reducing the resolution of the image and
and V are used to obtain the low-rank approximation of A as removing noise. The algorithm is widely used in various fields,
follows: including image compression, denoising, data analysis, and
image recognition. The results of the SVD algorithm can be

3
Authorized licensed use limited to: GITAM University. Downloaded on February 10,2025 at 05:44:40 UTC from IEEE Xplore. Restrictions apply.
improved by combining it with other techniques such as significant bits of either the singular values or singular vectors
independent component analysis, principal component are replaced with hidden data. The rationale behind this is that
analysis, and matrix factorization. The combination of LSB the singular values and vectors contain a significant amount of
and SVD can be used in a medical image processing project to information about the image, and small changes to them are
secure the information contained in medical images. less noticeable and do not significantly impact the visual
quality of the image.
b) LSB Steganography:
The LSB algorithm is a steganographic technique that a) Load an image and convert it to a grayscale matrix
involves replacing the least significant bits of the pixel values b) Perform SVD on the grayscale matrix to get the unitary
with bits from the hidden data to embed data within an image matrices U, D, and V and the singular values and
file. To implement the LSB algorithm, one can follow the vectors
following simple steps:
c) The least important bits of the single values or vectors
a) Load the image: Read the image file and load it into are changed to conceal secret data, which have a lower
memory as a matrix of pixel values. impact on the visual quality of the image, as they
b) Convert the data to be embedded into binary: The contain a large portion of the image information.
data to be embedded should be converted into binary d) Reconstruct the grayscale matrix using the modified
form. singular values or vectors
c) Embed the data: For each pixel in the image, replace e) Convert the grayscale matrix back to a colour image
the least significant bit of the binary data's red, green, and save it
and blue values as represented by bits.
d) Save the modified image: Save the modified image as This technique may not be very secure and could be easily
a new file. detected by steganalysis methods. More advanced techniques,
e) Decode the data: To retrieve the embedded data, you such as using singular value thresholding or using the singular
can decode the image by extracting the least vectors in a more sophisticated way, may be required to
significant bits from the green, red and blue values of achieve a higher level of security.
each pixel and converting the extracted bits back into
the original data. IV. EXPERIMENTAL RESULTS
It is important to note that the LSB algorithm will affect the
quality of the image, especially if a lot of data is embedded The proposed method, which involves combining LSB
within the image. The more data that is embedded, the greater steganography and AES encryption has been implemented
the degradation of the image quality. using MATLAB, with the addition of SVR compression to
ensure the preservation of image quality. A matrix is divided
Example: Consider the pixels 11111111 (255), which make into three smaller matrices using the mathematical approach
up the cover picture, and 11100001 (225), which make up the known as SVD, which represents the original matrix. In
secret image. In the event of an MSB substitution, 11101111 medical image hiding, SVD can be used to represent a medical
(239) will be the pixel for the stego image, but the outcome of image as a series of singular values, which can be manipulated
an LSB replacement will be 11111110. (254). to conceal information within the image.
The results of a medical image-hiding experiment using
SVD or LSB will depend on the specific implementation and
the quality of the data being used. In general, the results will
show the effectiveness of the method in hiding information
within the medical image while maintaining the quality of the
image and the accuracy of the information being hidden.
The image is a matrix of 512 by 512. Every pixel is
represented in the matrix by a number from 0 to 512. We are
looking for compression of that image using the SVD image
compression method. Using this method, we will retrieve some
information from the image but the image quality has to remain
good.

Fig. 4. LSB embedding SVD on the matrix A:

C. SVD+LSB: [ U, S, V,] = svd (A);

To combine SVD and LSB steganography, the image is
first decomposed into matrices using SVD, and then the least

4
Authorized licensed use limited to: GITAM University. Downloaded on February 10,2025 at 05:44:40 UTC from IEEE Xplore. Restrictions apply.
 The medical image-hiding experiment's results are
currently under evaluation and will be compared to the results
obtained using the LSB method without encryption. In the LSB
method, the image is embedded into a cover image. The
evaluation process will be based on the PSNR (Peak Signal-to-
Noise Ratio) and MSE (Mean Square Error) values, which can
be calculated using the following equations:
MSE is a metric used to measure the similarity or
difference between two images. It is calculated by taking the
difference between the pixel values of the original and the
processed image, squaring them, and taking their average. A
lower MSE value represents minimal distortion and greater
similarity between the two images. The formula for calculating
MSE is:
ଵ ଵ
MSE = ൈ σ஺஼ୀ଴ σ஻஽ୀ଴ (X(c, d) – y(c, d))2 (1)
஺ ஻
Where c and d are the numbers of rows and columns
respectively. A and B are the total numbers of rows and
columns, x is the cover image, and the Oswego image.
PSNR: PSNR is a metric used to measure image quality by
comparing the maximum possible power of a signal to the
amount of noise that corrupts the signal. The higher the PSNR
value, the better the image quality, as the signal is stronger
relative to the noise.
ெ஺௑೑
PSNR = 20 log10 ( ) (2)
ξெௌா

Fig. 5. IMAGE 1(SVD compression for different singular value “k”

Fig. 6. MAGE 2 (compression for different singular value “k” Fig. 8. Image A trasformation of cipher image, and cover image

Fig. 7. MAGE 3 (compression for different singular value “k”

TABLE I. SVD ALGORITHM RESULTS

IMAGE 1 IMAGE2 IMAGE3
K CR CI K CR CI K CR CI
5 175.5 160.5 5 176.5 160.8 5 177.3 161.5
10 88.9 150.6 10 83.7 151.2 10 84.5 152.2
100 7.8 170 100 7.9 170.99 100 8.1 171.5
Fig. 9. Image B trasformation of cipher image and cover image

5
Authorized licensed use limited to: GITAM University. Downloaded on February 10,2025 at 05:44:40 UTC from IEEE Xplore. Restrictions apply.
 cryptography and steganography. The SVD approach for
image compression can help to maintain image quality, while
the use of AES encryption can provide confidentiality for the
medical image. Furthermore, Using LSB steganography to
embed the encrypted medical image into a cover image can
help to conceal its presence and can add a further layer of
protection. It should be noted that the effectiveness of the
combined SVD and LSB steganography method in medical
image hiding relies heavily on the strength of the encryption
key and the robustness of the steganographic technique.
Common measures that evaluate how comparable the original
and reconstructed images are, including PSNR and MSE, can
be used to assess the reconstructed image's quality. Three
distinct kinds of medical images were used to assess the
performance of the suggested approach, and each image's
PSNR and MSE values were calculated.
Fig. 10. ImageC transformation of cipher image, and cover image
REFERENCES
TABLE II. PSNR AND MSE RESULTS [1] G. Prashanti, B. V. Jyothirmai, and K. S. Chandana, ‘Data
confidentiality using steganography and cryptographic techniques’, in
IMAGES PSNR MSE 2017 International Conference on Circuit, Power and Computing
A 71.567 0.00048 Technologies (ICCPCT), 2017, pp. 1–4.
B 62.476 0.00057 [2] R. Sharma and S. Bollavarapu, ‘Data security using compression and
C 62.567 0.00065 cryptography techniques’, International Journal of Computer
Applications, vol. 117, no. 14, 2015.
[3] A. Solichin and E. W. Ramadhan, ‘Enhancing data security using DES-
based cryptography and DCT-based steganography’, in 2017 3rd
International Conference on Science in Information Technology
(ICSITech), 2017, pp. 618–621.
[4] H. Al-Ghuraify, A. A. Al-Bakry, and A. T. Al-Jayashi, ‘Quaternion
Security Using Modifying Vernam Cipher With Image Steganography’,
The International Journal of Multimedia & Its Applications (IJMA) Vol,
vol. 11, 2019.
[5] M. A. Islam, M. A.-A. K. Riad, and T. S. Pias, ‘Enhancing security of
image steganography using visual cryptography’, in 2021 2nd
International Conference on Robotics, Electrical and Signal Processing
Techniques (ICREST), 2021, pp. 694–698.
[6] N. V. S. Reddy and Others, ‘Improving security in Image
Fig. 11. Graph for PSNR Steganography using MSB Bit differencing and Cryptographic
algorithm’, in 2018 International Conference on Computational
Techniques, Electronics and Mechanical Systems (CTEMS), 2018, pp.
This graph represents the model's image quality accuracy 228–230.
based on the PSNR matrix. [7] D. Bhat, V. Krithi, K. N. Manjunath, S. Prabhu, and A. Renuka,
‘Information hiding through dynamic text steganography and
cryptography: Computing and informatics’, in 2017 international
conference on advances in computing, communications and informatics
(ICACCI), 2017, pp. 1826–1831.
[8] K. M. Aishwarya, R. Ramesh, P. M. Sobarad, and V. Singh, ‘Lossy
image compression using SVD coding algorithm’, in 2016 International
Conference on Wireless Communications, Signal Processing and
Networking (WiSPNET), Chennai, India, 2016.
[9] P. Varghese and G. A. S. Saroja, ‘DWT, DCT and SVD Based
Hexagonal Image Compression’, in 2021 Fourth International
Conference on Electrical, Computer and Communication Technologies
(ICECCT), Erode, India, 2021.
[10] R. Ranjan, P. Kumar, K. Naik, and V. K. Singh, ‘The HAAR-the JPEG
based image compression technique using singular values
decomposition’, in 2022 2nd International Conference on Emerging
Fig. 12. Graph for MSE Frontiers in Electrical and Electronic Technologies (ICEFEET), Patna,
India, 2022.
This graph represents the model's image quality accuracy [11] A. H. Bentbib, K. Kreit, and I. Labaali, ‘Randomized tensor singular
based on the MSE matrix. value decomposition for multidimensional data compression’, in 2022
11th International Symposium on Signal, Image, Video and
V. CONCLUSION Communications (ISIVC), El Jadida, Morocco, 2022.
The proposed method seems to provide a comprehensive [12] M. A. Usman and M. R. Usman, ‘Using image steganography for
providing enhanced medical data security’, in 2018 15th IEEE Annual
approach to securing medical images by combining

6
Authorized licensed use limited to: GITAM University. Downloaded on February 10,2025 at 05:44:40 UTC from IEEE Xplore. Restrictions apply.
 Consumer Communications & Networking Conference (CCNC), Las using 512-bit algorithm with LSB technique’, in 2021 5th International
Vegas, NV, 2018. Conference on Electrical Information and Communication Technology
[13] S. S. Yadahalli, S. Rege, and R. Sonkusare, ‘Implementation and (EICT), Khulna, Bangladesh, 2021.
analysis of image steganography using Least Significant Bit and [15] P. Subhasri, S. Santhoshkumar, A. Sumathi, C. Balakrishnan, and F.
Discrete Wavelet Transform techniques’, in 2020 5th International Kurus Malai Selvi, ‘Efficiency enhancement using least significant bits
Conference on Communication and Electronics Systems (ICCES), method in image steganography’, in 2022 International Conference on
Coimbatore, India, 2020. Automation, Computing and Renewable Systems (ICACRS),
[14] N. Nahar, M. K. Ahmed, T. Miah, S. Alam, K. M. Rahman, and M. A. Pudukkottai, India, 2022.
Rabbi, ‘Implementation of android based text to image steganography

7
Authorized licensed use limited to: GITAM University. Downloaded on February 10,2025 at 05:44:40 UTC from IEEE Xplore. Restrictions apply.