Santoso 2019 j._phys.__conf._ser._1175_012057

Journal of Physics: Conference Series
PAPER • OPEN ACCESS
Color-based microscopic image steganography for telemedicine
applications using pixel value differencing algorithm
To cite this article: Budi Santoso 2019 J. Phys.: Conf. Ser. 1175 012057
View the article online for updates and enhancements.
This content was downloaded from IP address 180.246.243.172 on 22/10/2020 at 23:32

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd
1st International Conference on Advance and Scientific Innovation (ICASI)
IOP Conf. Series: Journal of Physics: Conf. Series 1175(2019) 012057
IOP Publishing
doi:10.1088/1742-6596/1175/1/012057
1
Color-based microscopic image steganography for
telemedicine applications using pixel value differencing
algorithm
Budi Santoso
Engineering Faculty, Dr.Soetomo University, Semolowaru 84th
Surabaya, Indonesia
*budi.santoso@unitomo.ac.id
Abstract. There are some ethics in the medical world to keep patient information confidential
on every result of medical observations and analysis. Therefore, steganography technique is an
alternative to conceal sensitive text information on a medical image in the telemedicine system
process. Research on steganography using Simple LSB-Substitution and other algorithm has
been available but in this study, the implementation of steganography was designed to hide text
in color medical images before being sent through the telemedicine system. Sensitive text is
hidden in medical images using the Pixel Value Differencing (PVD) algorithm. The main
purpose of using this algorithm is to maintain the peak signal to noise ratio (PSNR) and mean
square error (MSE) of the steganography image result. As a result, the PSNR value can be
maintained at the 57,98dB for 10KB hidden text, and lower MSE at 0.05 for high-density
object microscopic image.
1. Introduction
The confidentiality of information, especially the personal data of patients and the results of medical
diagnosis, must be maintained, in accordance with the code of ethics in the medical world. Meanwhile,
the development of communication technology has made possible the process of diagnosis and remote
medical observation, where patients and medical experts are not in one place - or commonly referred
to as telemedicine. Because of the range of distance between patients and medical experts, the process
of transmitting data from diagnoses or observations through media that are already available in
communication technology is needed[1]. Media commonly used in data transmission processes are
electromagnetic waves (radio propagation) and internet networks. However, the transmission through
both media requires an additional process so the data cannot be tapped and only reaches the specified
destination.
There are many ways to secure data that can be done in the data transmission process, such as
watermarking[2][3][4], cryptography[5][6], and encryption[7][8]. But steganography is one of the
more appropriate ways to do because it can send two data formats both images and text, and also
simultaneously encoded. According to T.Morkel et.al[9], steganography is defined as the science and
art of hiding secret messages (hiding messages) so that the existence of messages is not detected by
the human senses. The media used is generally a different medium from the secret information carrier
media, this is the function of steganography as a disguise technique using different media to hide
confidential information so it is not clearly visible.

IOP Publishing
doi:10.1088/1742-6596/1175/1/012057
2
1.1. Pixel Value Differencing
Pixel Value Differencing (PVD) is one method used in steganography and operates in the spatial
domain. Early research on PVD algorithms was carried out by Wu et al., 2001, by looking for
differences in the value of the two closest pixels. The difference is used to determine the amount of
data that can be inserted based on the range of the selected table.
Based on an analysis of the human visual system which states that the human eye is not sensitive to
changes in pixels that have a high contrast but sensitive to changes in pixels that have low contrast.
Through these properties, more bits of confidential data can be inserted in pixels that have high
contrast values, and fewer bits that can be inserted in pixels with low contrast. This is the basis of the
PVD algorithm on steganography.
The PVD algorithm is a system of two-pixel closest search direction as shown in Figure 1:
Figure 1. PVD algorithm concept
The insertion process in this method is done by comparing two neighboring pixels Pi and Pi+1 using
equation (1).
𝒅𝒊 = |𝑷(𝒊) − 𝑷(𝒊+𝟏)| (1)
The results of these comparisons are used to find out how many bits can be inserted into the two
compared pixels. This method uses the Wu and Tsai schemes to determine the range of the previous
pixel comparison. The Wu and Tsai schemes used are R = {[0-7], [8-15], [16-31], [32-63], [64-127],
[128-255]}.
Table 1. A The Wu and Tsai R schemes
Range 0-7 8-15 16-31 32-63 64-127 128-255
Hiding
Capacity 3 3 4 5 6 7
This scheme is used to find out which range is the difference between the two pixels if it is known
where the range is, then we can find the lower limit (U𝒊) and upper limit (L𝒊). After knowing the lower
limit and upper limit of the range, we can calculate the width of the optimum range (Wi) using
equation (2).
𝑊𝑖 = 𝑈𝑖 − 𝐿𝑖 + 1 (2)
The next step is to find the number of message bits (t𝒊) that can be inserted through equation (2.3).
𝑡𝑖 = [log2(𝑊𝑖)] (3)
Message insertion can be done by taking as many as 𝒊 bits of the message to be inserted. After
knowing how many message bits will be inserted, change the message bits to be inserted into decimal
(). Then the new difference value is calculated for insertion into the image using equation (4).
𝑑′𝑖 = 𝑏 + 𝐿𝑖 (4)
To determine the value of new pixels that have been inserted messages, there are several rules that
must be met with equation 5:

IOP Publishing
doi:10.1088/1742-6596/1175/1/012057
3
(𝑃′
(𝑖), 𝑃′
(𝑖+1)) =
{
(𝑃(𝑖) + [
𝑚
2
], 𝑃(𝑖+1) − [
𝑚
2
]) , 𝑖𝑓 𝑃(𝑖) ≥ 𝑃(𝑖+1) 𝑎𝑛𝑑 𝑑′
𝑖 > 𝑑𝑖
(𝑃(𝑖) − [
𝑚
2
], 𝑃(𝑖+1) + [
𝑚
2
]) , 𝑖𝑓 𝑃(𝑖) < 𝑃(𝑖+1) 𝑎𝑛𝑑 𝑑′
𝑖 > 𝑑𝑖
(𝑃(𝑖) − [
𝑚
2
], 𝑃(𝑖+1) + [
𝑚
2
]) , 𝑖𝑓 𝑃(𝑖) ≥ 𝑃(𝑖+1) 𝑎𝑛𝑑 𝑑′
𝑖 ≤ 𝑑𝑖
(𝑃(𝑖) + [
𝑚
2
], 𝑃(𝑖+1) − [
𝑚
2
]) , 𝑖𝑓 𝑃(𝑖) < 𝑃(𝑖+1) 𝑎𝑛𝑑 𝑑′
𝑖 ≤ 𝑑𝑖}
Where,
𝑚 = 𝑑′𝑖 − 𝑑 (5)
These processes are carried out continuously until the message bits are all inserted into the image.
The process of extracting messages from the stego image using the PVD algorithm begins by sorting
all the pixels in the image that has been inserted into the message, according to the way the message is
retrieved. Then the difference in the value of the new difference value (𝒊) is calculated. The new
difference value is used to determine the continuous ranges (R) value that has been defined using the
Wu and Tsai schemes. Thus the message that has been inserted () is obtained. Based on this
information it can be seen how long the secret data is inserted in both pixels so the secret message that
has been inserted is retrieved.
1.2. Least Significant Bit (LSB) Steganography
Least significant bit (LSB) insertion is a simple approach to embedding text in a cover image[10]s.
The least significant bit or the 8th bit from some of the bytes inside an image is changed to a bit of the
secret message. When using a color image, a bit of each of red, green and blue color components can
be used, since they are each represented by a byte. In this study, steganography using LSB insertion is
used as a comparison of the of the PVD steganography algorithm image output. The results of the
PSNR comparison of the two methods are presented in the form of tables and graphs.
2. Result and Discussion
In this study 2 groups of test data were used from the ZN Sputum Smear Microscopy Image Database
(ZNSM-iDB). The first group contained 20 microscopic images with 1-10 bacilli (low density bacili
image) as a result of the Zielh Nielsen staining process. Whereas the second group contained 20
microscopic images with many bundled bacilli (high density bacili image) resulting from the Zielh
Nielsen staining process.
Figure 2. Low objet density
microscopic image
Figure 3. High objet density
microscopic image
In each image the process of hiding data is done with text files size of 10KB, 20KB and 30KB.
The original image is treated as a cover image at the steganography process, both for the PVD method
and the LSB method as a comparison. In each test calculation of each MSE and PSNR is carried out,
as the output image quality parameters for steganography.

IOP Publishing
doi:10.1088/1742-6596/1175/1/012057
4
2.1. Test Results of High-Density Objects (HDO)
Table 4 shows the overall results of steganography using PVD algorithm experiments for groups of
images with high object density both for MSE and PSNR measurement.
Table 2. Results of steganography using PVD for high-density object
image
PSNR (dB) MSE
text
10KB
text
20KB
text
30KB
text
10KB
text
20KB
text
30KB
01 58.0525 54.9574 53.2387 0.0527 0.1074 0.1597
02 58.3286 55.3168 53.5693 0.0491 0.0982 0.147
03 58.0522 55.077 53.3029 0.0526 0.1044 0.1572
04 58.2875 55.083 53.402 0.0496 0.1037 0.1528
05 57.7827 54.7891 53.1601 0.0553 0.1106 0.1616
06 57.9707 55.1059 53.4114 0.053 0.1029 0.1521
07 57.813 54.8026 53.0383 0.0562 0.1122 0.1685
08 58.339 55.321 53.5255 0.0492 0.0986 0.149
09 58.2085 55.2601 53.5348 0.0506 0.0994 0.148
10 58.0876 55.0981 53.4004 0.052 0.1034 0.153
11 58.2389 55.0726 53.3555 0.0503 0.1035 0.1538
12 57.8192 54.8207 53.0956 0.0561 0.1118 0.1661
13 58.041 54.9614 53.1687 0.0524 0.1064 0.1609
14 58.1478 55.0164 53.1872 0.0512 0.1049 0.1599
15 56.4596 53.5155 51.7487 0.0842 0.1648 0.2477
16 58.2181 55.1944 53.3865 0.0508 0.1022 0.1553
17 58.0486 55.1243 53.3479 0.0527 0.1037 0.1564
18 57.8385 54.8681 53.1697 0.0547 0.1084 0.1607
19 57.7224 54.7427 53.0455 0.0575 0.1141 0.1683
20 58.244 55.2506 53.4722 0.0501 0.0998 0.1502
2.2. Test Results of Low-Density Objects (LDO)
Table 3 shows the overall results of experiments for groups of images with low object density both for
MSE and PSNR measurement.
Table 3. Results of steganography using PVD for low-density object
image
PSNR (dB) MSE
text
10KB
text
20KB
text
30KB
text
10KB
text
20KB
text
30KB
01 57.3763 54.3879 52.7102 0.0628 0.125 0.1835
02 57.5083 54.5336 52.7643 0.06 0.1193 0.1793
03 57.5004 54.4378 52.6857 0.0606 0.1223 0.1831
04 57.9232 54.9458 53.1905 0.0546 0.1083 0.1622
05 58.1048 55.1562 53.3953 0.0521 0.1028 0.1542
06 57.5526 54.5602 52.8391 0.0596 0.1189 0.1768
07 57.9757 54.8825 53.0688 0.0539 0.1102 0.1677
08 57.3742 54.3887 52.6122 0.0628 0.1252 0.1884
09 57.2537 54.2082 52.4063 0.0649 0.1311 0.1987

IOP Publishing
doi:10.1088/1742-6596/1175/1/012057
5
10 57.7416 54.7735 53.0174 0.0571 0.1129 0.1691
11 57.9162 54.7335 52.9255 0.0549 0.1151 0.1747
12 57.3804 54.2713 52.4613 0.063 0.1294 0.1969
13 57.8996 54.9183 53.1964 0.0549 0.109 0.1619
14 57.764 54.8299 53.0685 0.0569 0.1117 0.1673
15 57.8527 55.0402 53.3048 0.0541 0.1059 0.1576
16 57.5291 54.4319 52.5515 0.0603 0.1238 0.192
17 58.0589 55.0337 53.2567 0.0526 0.1058 0.1596
18 57.7674 54.8931 53.1683 0.056 0.1091 0.1625
19 57.3452 54.4089 52.6515 0.0634 0.1244 0.1865
20 57.3874 54.3969 52.667 0.063 0.1252 0.186
From the test results, it was found that the average MSE and PSNR values in the PVD
steganography process are better than the results obtained from the LSB steganography method.
Whereas in testing the type of microscopic image groups, it was found that the group of images with a
low bacilli density had higher MSE values and a lower PSNR. This is advantageous because in a
microscopic image with higher bacilli density requires more observation accuracy as one of the basic
diagnoses by the medical expert.
Figure 4. Graph of PSNR PVD Steganography results on HDO
Figure 5. Graph of PSNR PVD Steganography results on LDO
48
50
52
54
56
58
60
1 3 5 7 9 11 13 15 17 19
PSNR(dB)
10kb-txt
20kb-txt
30kb-txt
49
50
51
52
53
54
55
56
57
58
59
1 3 5 7 9 11 13 15 17 19
PSNR(dB)
10kb-txt
20kb-txt
30kb-txt

IOP Publishing
doi:10.1088/1742-6596/1175/1/012057
6
Generally, -both the PVD and LSB methods- the PSNR value decreases when the size of the
message stored in the image increases as shown at Table 4.
Table 4. Average results of MSE and PSNR steganography using PVD
Average of MSE
10kb text 20kb text 30kb text
PVD Method (HDO) 0.054015 0.10802 0.16141
PVD Method (LDO) 0.058375 0.11677 0.1754
LSB Method (HDO) 0.32186 0.64036 0.956065
LSB Method (LDO) 0.32664 0.64282 0.96246
Average of PSNR (dB)
10kb text 20kb text 30kb text
PVD Method (HDO) 57.98502 54.96889 53.22805
PVD Method (LDO) 57.66059 54.66161 52.89707
LSB Method (HDO) 50.09723 47.12932 45.40128
LSB Method (LDO) 50.149575 47.19772 45.4311
3. Conclussion
Based on the results of steganography testing for microscopic color images using the PVD algorithm,
it is found that the PVD algorithm produces a stegano image with better PSNR than the comparison
method (LSB). The best average value is 57.98 dB in the 10KB text. Besides that, it is found that the
MSE measurement results on PVD algorithm steganography output with high object density images
are even lower, that is 0.05 in the 10KB text.
References
[1] Sachpazidis D I, 2008 Image and Medical Data Communication Protocols for Telemedicine and
Teleradiology September.
[2] Zain J and Clarke M, 2005 Security in Telemedicine : Issues in Watermarking Medical Images
in 3rd International Conference: Sciences of Electronic,Technologies of Information and
Telecommunications.
[3] Eswaraiah R and Reddy E S, 2015 Robust medical image watermarking technique for accurate
detection of tampers inside region of interest and recovering original region of interest 2 p. 615–
625.
[4] Eswaraiah R and Reddy E S, 2014 Medical Image Watermarking Technique for Accurate
Tamper Detection in ROI and Exact Recovery of ROI Int. J. Telemed. Appl. 2014.
[5] Al-haj A Abandah G and Hussein N, 2015 Crypto-based algorithms for secured medical image
transmission IET Inf. Secur. 9 p. 365–373.
[6] Varsha and Chhillar R S, 2015 Data Hiding Using Steganography and Cryptography Int. J.
Comput. Sci. Mob. Comput. 4, 4 p. 802–805.
[7] Syawal M F Fikriansyah D C and Agani N, 2016 Implementasi Teknik Steganografi
Menggunakan Algoritma Vigenere Cipher Dan Metode LSB 4, 3.
[8] Rahim R, Pixel Image Steganography Using EOF Method and Modular Multiplication Block
Cipher Algorithm.
[9] Morkel T Ellof J H P and Olivier M ., 2005 An overview of image steganography in
Proceedings of the Fifth Annual Information Security South Africa Conference (ISSA2005) July
2005.
[10] Champakamala B S Padmini K Professors R D K A and Bosco D, 2014 Least Significant Bit
algorithm for image steganography Overview of Steganography Int. J. Adv. Comput. Technol. 3,
4 p. 5.

Santoso 2019 j._phys.__conf._ser._1175_012057

Recommended

Recommended

More Related Content

What's hot

What's hot (19)

Similar to Santoso 2019 j._phys.__conf._ser._1175_012057

Similar to Santoso 2019 j._phys.__conf._ser._1175_012057 (20)

Recently uploaded

Recently uploaded (20)

Santoso 2019 j._phys.__conf._ser._1175_012057