Prema T.

Akkasaligar and Sumangala Biradar

BLDEA’S

V.P. Dr. P. G. H. College of Engineering and Technology, India

Abstract

Nowadays

due to rapid growth in the medical field, transmission of digital medical

images over internet has become frequent. Due to lack of security levels in

internet intruders can access the digital medical images and alter the digital

medical images. Ensuring the security of

digital medical image has become an important issue. In this regard several

image encryption techniques based on deoxyribonucleic acid (DNA) computation

and chao’s theory is already available. But these methods take more

computational time and also digital medical image require high-level security.

To enhance and provide high-level security for digital medical image, multilevel

security is proposed using Lorenz chaotic map and DNA sequences. Also the

selective digital medical image encryption algorithm is proposed to reduce

computational time. In the proposed method, firstly the input digital medical

image is divided into blocks. The selected pixels in blocks are renovated into

DNA encoded matrix. Later, chaotic sequence generated by Lorenz chaotic map is

used to scramble the selected pixels positions of DNA encoded matrix. Finally

DNA add operation is used for the fusion of scrambled DNA encoded matrix blocks

to get an encrypted digital medical image. The decryption is inverse process of

encryption and DNA sub operations in place of DNA add operation. The experimental

result shows that proposed algorithm not only enhances security level but also

reduces computational time. And also encumber exhaustive and statistical

attacks.

Keywords:

Digital Medical Image, Encryption, Decryption, Lorenz chaotic map, DNA

sequences

INTRODUCTION

We

are living in the automation world where everything will be digitalized. In

medical field e-health, smart health and telemedicine are automated systems. These

systems utilize advanced digital

communication system to transfer and receive medical information for end-to-end

communication. These digitalization reduce time and remote centric. However,

due to the open source, the transmission of digital medical image can easily

intercepted by hackers. Hence image encryption technology has a prime concern

in present era. In image encryption the digital image is encrypted then it

sends via insecure channel. At receiver end only authorized user can access the

original image. In this field several cryptographic techniques are available

among those chaotic system and DNA sequence based techniques are advanced

techniques.

The chaotic system generates a nonlinear deterministic

pseudo random sequence, which are highly sensitive to initial conditions. It

also has many characteristics, such as, ergodicity, mixing property and

structural complexity. The Lorenz chaos system is high dimensional chaotic map

and it is very complex. Due to these properties, Lorenz chaotic system is used

in the field of cryptography for digital medical image encryption. But because

of limited key space it can easily attacked by the key space and key sensitive

analysis test. Hence new technique called DNA technique evolved in

cryptography.

The DNA cryptography is a fresh domain in

cryptography. DNA is used as an information carrier in image encryption. DNA

computing uses biological DNA molecules as computing medium and biochemical

reaction as computing tools 1. The main security is based on DNA structure. The

implementation of DNA computation on image is not easy because of complex DNA

structure. Hence it is used only to encrypt the character information. For image encryption only DNA sequence is

used as security which is not enough. To overcome from the above drawbacks, the

combination of Lorenz chaos system with DNA sequence is used for image

encryption.”

Background

A slight work has been

done by DNA cryptography and chaotic system to transmit a digital medical image

in a secured way. Several encryptions of images had been put forward by using cryptographic

and watermarking techniques.

In 2, authors have

proposed cosine number transform (CNT) method for medical image encryption. It

is very sensitive to changes, which is suitable for medical image. The image is

divided into blocks and CNT method is applied twice to encrypt the image. In 3,

authors have presented 2D chaotic map to permute the pixel of the medical

image. The image is divide into blocks, sensitive part of the blocks are

identified and masked with synthetic image to encrypt the selective part of the

medical image. In 4, authors have proposed digital watermarking method. The

entropy and mean of the medical image is calculated to obtain cipher image. The

cipher image is hidden using digital watermarking algorithm to get encrypted

medical image. In 5, authors have proposed the Rivest, Adi Shamir and Leonard

Adleman (RSA) algorithm for encryption and decryption of magnetic resonance

imaging (MRI) images. Further to extort the tumor details K-means, watershed segmentation

is used. In 6, authors proposed least significant bit (LSB) embedding

algorithm to embed patient privacy

information in the high frequency components of a transformed image. Next the

data concealed image is encrypted using Linde, Buzo and Gray (LBG) algorithm to

ensure security. In 7, the electronic

code book (ECB) mode of advanced encryption standard (AES) is used for

encryption of patient record. The image is divided into region of interest

(ROI) and region of non interest (RONI). The discrete wavelet transform (DWT)

and inverse discrete wavelet transform (IDWT) are used for embeddin g encrypted

patient record in the RONI region.

In 8, the wavelet domain

watermarking technique is used to embed the Electronic health record in to the medical

image, there by generating a watermarked medical image. In the second stage of

the algorithm, a number of deoxyribonucleic acid masks are created using

logistic map function and DNA conversion rules. Then encryption is performed on

the watermarked image to generate a number of cipher images. Genetic algorithm

(GA) is applied to find the best DNA mask in iterative manner until the

condition is met.

Main FOCUS OF the CHAPTER

Issues, Controversies, Problems(Subhead 2:

Arial, Size 12, Title Case, Bold)

The digital medical image is usually very large in

size and contains very sensitive, confidential data. Providing the security to

the digital medical image and transmitting in less time is a big challenge. The

main aim of the research work contains design and development of efficient

methods to enhance the security level and improve the performance of the time

efficiency.

The ancient techniques cannot survive every possible attack

and not robust. The DNA cryptography fulfills the requirements of digital

medical image due to its uniqueness. To enhance and provide high-level security

for digital medical image, multilevel security is proposed using Lorenz chaotic

map and DNA sequences.

BASIC

OPERATIONS

Lorenz Chaotic System

The Lorenz chaotic map is coined with butterfly effect

and very sensitive to initial conditions. The chaotic sequences generated are a

high dimensional and periodic. The sequences are very sensitive to initial

conditions and more complex. Hence it is suitable for digital medical images to

provide security and confidentiality. The Lorenz chaotic

system is defined in (1)-(3).

P1=

?(Q?P) (1)

Q1=

rP?Q?PR

(2)

R1=

PQ?bR

(3)

where P, Q and R are arbitrary

parameters and ?=10, r=28 , b=8/3 are positive constants.

DNA Sequences

The basic elements of biological DNA are nucleotide,

because of the different chemical structure, nucleotide are divided into four

basic alphabets namely, Adenine (A), Guanine (G), Cytosine (C) and thymine (T).

Owing to the key hydrogen, the two chains are put together, and form a double

helix structure chain and that one chain in the base sequence complementary to

the other, that is, A and T are pairs , G and C are pairs as shown in Fig.1.

In the binary, 0 and 1 are complementary, therefore 00

and 11 are complementary, 01 and 10 are complementary. So that A, T, G and C

nucleic acid bases can be encoded as 00, 11, 10 and 01 respectively. Using this

concept we can get 4! = 24 different encoding patterns. But due to the

Table 1. Eight kinds of DNA

map rules

1

2

3

4

5

6

7

8

A

00

00

01

01

10

10

11

11

T

11

11

10

10

01

01

00

00

G

01

10

00

11

00

11

01

10

C

10

01

11

00

11

00

10

01

complementary relation between DNA bases only eight

patterns of encoding satisfy the complementary base pairing shown in Table 1.

PROPOSED

METHODOLOGY

Digital medical image encryption is performed by

combining the Lornenz chaotic system and DNA operation. To enhance the security level in the proposed

selective medical image encryption algorithm multilevel encryption is used. In

the proposed model in first phase, the grayscale input digital medical image is

converted into an 8-bit binary image. The DNA sequence as A=01, T=10, G=11and

C=00 is applied on 8-bit binary image. The DNA encoded matrix is obtained. The

DNA encoded matrix is divided into 8 blocks and random pixels are selected. In

the second phase Lornez chaotic map with state variables and control parameters

are used to generate the chaotic sequence. The chaotic sequence is sorted and

based on index of the sorted chaotic sequence the selected pixels of the DNA

encoded matrix blocks are scrambled. In the third phase DNA add logical operation

is performed for the fusion of the scrambled DNA encoded matrix blocks to get

the intermediate encrypted image. The Table 2 shows the DNA add operation used

in this scheme. From Table 2 it is clear that the results of DNA add operation

is unique. The intermediate encrypted image is decoded using DNA sequence to

get final cipher image. The cipher image is decrypted using inverse process of

encryption and DNA sub operations as shown in Table 3 in place of DNA add

operation.

Table 2. DNA Add Operation

ADD

T

A

C

G

T

C

G

T

A

A

G

C

A

T

C

T

A

C

G

G

A

T

G

C

Table 3. DNA Sub Operation

SUB

T

A

C

G

T

C

G

T

A

A

A

C

G

T

C

T

A

C

G

G

G

T

A

C

The detailed steps

of selective digital medical image encryption are shown in Figure 2.

The selective

digital medical image encryption process

is represented in Algorithm.1

Algorithm 1. Selective Digital Medical Image Encryption

(SDMIE)

Step 1: Start

Step 2: The grayscale input image is represented as I(m, n),

where m is the row size and n is

column size. Further, it is converted

into 8 bit binary image matrix of size m rows and n×8 columns

Step 3:

The binary image is renovated into DNA encoded matrix as D(m, n) of size m rows and n×4 columns

Step 4: The DNA encoded matrix as D(m, n) divided into 8 blocks

Step5: The chaotic

sequences X,Y are generated using Lornez chaotic map. The chaotic sequences are

rearranged in increasing order as X1, Y1 to alter the pixel values

Step 6: The index value of

X1 and Y1 chaotic sequences are

used to scramble the randomly selected pixels of D(m, n) blocks .The

scrambled pixels of the matrices are

represented as D1(m ,n)

Step 7: DNA add operation is used for the fusion of D1(m ,n)

blocks. The result of fusion is denoted as G(m ,n)

Step

8: Now transform the matrix G(m, n) into decimal using Step 2 inversely and encrypted digital medical image E(m ,n) is

obtained. This is the ciphered image

Step 9: Stop

The decryption is performed using inverse

process of SDMIE algorithm and DNA sub operation.

PERFORMANCE ANALYSIS

A image

encryption algorithm has to resist all types of security analysis like

statistical attacks, differential attacks and exhaustive attacks. For

statistical attack, histogram analysis and correlation coefficient analysis are

used. The number of changing pixel rate (NPCR) and unified average changed

intensity (UACI) are used to check differential attack. The key space analysis

and key security analysis are used for exhaustive attack. The mean square error

(MSE) and peak signal to noise ratio (PSNR) are used to check the quality of encrypted

image.

Histogram Analysis

The histogram

analysis shows the distribution of pixel value based on intensity. For

encrypted image based on the scatter of pixel value the cryptanalysis judge the

strength of image encryption algorithm.

Correlation Coefficient Analysis

The

correlation coefficient is measure of correlation between the neighboring

pixels in the given images. The good encryption algorithm must have highly

correlated adjacent pixels. The Pearson’s correlation coefficient is shown in

eq. (4).

(4)

where I1

and I2 are the grayscale values of the input digital medical image and encrypted digital

medical image. The N is size of the image. The value of r varies between +1 and

-1. A value equal to zero means no correlation 9.

NPCR and UACI

The NPCR and

UACI are two decisive factor used to measure the performance of image

encryption methods against the differential attacks. The NPCR and UACI are

defined in eq.(5) and eq.(7) respectively.

(5)

where W1 and H1 are

width and height of the image and D1(i1,j1) is

defined as

(6)

(7)

where I1 and

I2 are two ciphered images respectively obtained from original image

and one pixel value changed original image.

MSE and PSNR

The MSE and PSNR

are two metrics used to check whether the alteration of noise or error effects

the quality of the image. The MSE estimates the average of squares of the

errors between the input digital medical image ‘I1’ and encrypted digital medical image ‘ I2 ‘9.

The MSE and PSNR are defined in eq. (8) and eq.(9) respectively.

(8)

(9)

where N

represents size of the image.

EXPERIMENTAL RESULTS

The experimentation is carried out on digital medical

image. The Matlab (R2012a) software is used to implement the proposed method.

The Figure 2(a) shows the sample input digital medical image. In the proposed

SDMIE algorithm DNA map rule-2 as specified in Table 1 used to attain DNA

encoded odd matrix. In Lornez chaotic map the initial values of arbitrary parameters

are P=1.2, Q=1.2 and M=3.7 are considered to generate Lorenz chaotic sequence.

The Lorenz chaotic sequences are sorted and position of the sorted sequences is

used to scramble the randomly selected pixels of DNA encoded matrix. The ADD

operation specified in the Table 2 is used for the fusion of scrambled matrix to

get an encrypted digital medical image as shown in Figure 2 (b).

The performance analysis demonstrates that the

proposed algorithm provides high security. The Figure 3(a) shows the histograms

of the input digital medical image, Figure 3(b) shows encrypted digital medical

image and Figure 3(c) shows decrypted digital medical image. The histogram

analysis shows that the encrypted digital medical image

pixels are changed and it is equally distributed.

Key Space Analysis

In the proposed

SDMIE algorithm the primary values of state variables and the system parameters

of Lorenz chaotic maps are used as secret key. Thus, there are six secret keys

(K, L, M, ? , r, b) are used in proposed algorithm. The key size is 1011×1011×1011×1011×1011× 1011=1066, if

the precision is 1011. The secret key space is very huge to verify

exhaustive attack.

Key Security Analysis

The

Lorenz chaotic system is highly very sensitive to initial conditions of state

variables and control parameters. If

there is a slight modification then retrieving same input digital medical image

from decryption process is impossible. The secret key test is shown in Figure

4(a), where digital medical image is decrypted with wrong key q0=0.00000001

instead of q0=1 and histogram of decrypted digital medical image

with wrong key is shown in Figure 4(b). The Figure 4 (a and b) shows that the

decrypted digital medical image is not same as the input digital medical image

and the histogram of the decrypted digital medical image is comparatively

consistent. The other parameters (secret keys) are also very sensitive.

Hence proposed methodology is very sensitive to the keys. It shows that

SDMIE algorithm resist against the exhaustive attack.