A novel encryption method for medical images using 2D Zaslavski map and DNA cryptography
- 99 Downloads
Medical imaging describes the noninvasive methods that enable medical practitioners to look inside the body, which are critical in modern clinical diagnosis. The diagnosed medical images may have confidential information related to patient’s privacy. Privacy and security must be guaranteed for the digital images over Internet, to ensure its confidentiality. In this work, a novel encryption method is suggested for medical images. The work is intended to suggest a novel encryption scheme for medical images which is strong, secure and efficient. Also, it can be applied to any kind of the medical images, irrespective of its storage format. In the proposed method, the pixels of the image are shuffled using a pseudorandom number generator based on two-dimensional Zaslavski map. The permuted image is encrypted by DNA encryption. Various visual analysis, correlation analysis, quality and security analysis techniques are applied to verify the performance of the method.
KeywordsMedical image encryption Chaotic map Zaslavski map DNA cryptography Histogram analysis
Because of the rapid advancement of multimedia and communication technology, digital images are often used for communication. Medical images include radiography, computed tomography (CT), magnetic resonance imaging (MRI), photoacoustic imaging, etc. A patient’s medical image may include his medical records along with text-based personal information, clinical diagnosis and examination records . With the recent developments in technology, medical field through E-health, or telemedicine is popular. In health sector, medical imaging has become a major part of most diagnostic procedures. Medical images can inform the doctor about the internal problems which could not be detected by external assessment of the patient. Unfortunately, traditional encryption algorithms such as DES, AES, etc., are not well suited for medical images due to their large size and different storage formats. Many encryption techniques exist to overcome the spying of information in Internet and to assure protection of patient’s private information.
Xiao Chen and Chun-Jie Hu suggested an ‘Adaptive medical image encryption algorithm based on multiple chaotic mapping’ . In this work, multiple chaotic maps are applied in permutation phase and XOR operation for diffusion. Bhasker Mondal et al. describe a light weight secure encryption scheme based on chaos and DNA computing for grayscale images, in which encryption is implemented using 1D logistic map . Dridi Manal and Mtibaa Abdellatif explain the encryption of medical images using DCT and two-dimensional Arnold Cat map in the paper ‘crypto-compression of medical image based on DCT and chaotic system’ .
In this work, we use an encryption algorithm which has two phases, pixel permutation using chaotic map and diffusion using DNA cryptography. Many of the recent works show that chaotic map is appropriate to generate pseudorandom sequences. Chaos-based cryptosystems are suitable for image encryption owing to the inherent properties of chaotic systems such as simple structure, ergodicity and high sensitivity to initial values and control parameters . In the permutation phase, we have used a pseudorandom number generator based on two-dimensional Zaslavski map . Diffusion phase makes use of DNA encryption in order to improve security of the system.
The rest of this paper is organized as follows. Section II describes the properties of chaotic system and the pros of 2D Zaslavski map. The encryption method is explained in Section III. The experimental results are shown in Section IV. Section V shows the performance analysis of the algorithm using various tests. Finally, conclusion is drawn in section VI.
2 Chaotic map in cryptography
Chaos is subtle nonlinear dynamical systems that are highly sensitive to initial conditions. The possibility for self-synchronization of chaotic oscillations has sparked an avalanche of works on application of chaos in cryptography . The strong mixing property, sensitive dependence on initial conditions, continuous broadband power spectrum and control parameters make chaotic map advisable for cryptographic primitives. As chaotic maps are non-period, non-converging, discrete-time dynamical systems, they can be used to generate pseudorandom series. Also, because chaotic maps are cryptographically secure random number generators, they can be good enough to be used in image encryption.
2.1 2D Zaslavski map
In the diffusion phase, we applied the principles of DNA encryption. DNA is a chain of nucleic acids: Adenine(A), Thymine(T), Guanine(G), Cytosine(C). A pixel of the image can be represented as the combination of four nucleic acids using DNA digital coding  patterns. It is possible to write eight DNA coding patterns, which satisfies the biological complementarity of bases.
Each pixel of the permuted image is represented with DNA patterns by randomly selecting one of the eight patterns. From the biomedical and genomic database NCBI, one of the considerably large DNA sequences is selected for key generation. NCBI is the major resource of bioinformatics tools and services. For our experimental simulation, a DNA sequence with more than 400,000 bases is accessed from the database. N bases from the accessed DNA sequence are taken randomly to form the key for encryption, since the DNA representation of the image also has N bases.
DNA addition rules
DNA subtraction rules
4 Simulation result
5 Performance analysis
The performance of the proposed method is tested and analyzed with various metrics as shown below.
5.1 Randomness tests
Randomness test results
Value of P
Frequency monobit test
Frequency test in a block
Binary rank test
5.2 Security analysis
5.2.1 Histogram analysis
Histogram of the encrypted image must be totally different from that of original image . Since a good image encryption method tends to encrypt a plain image to random-like, it is desired to see a uniformly distributed histogram for a cipher image . Hence, the method is found to be secure.
5.2.2 Histogram deviation
Histogram deviation  measures the deviation between histograms of original and encrypted images to ensure the quality of encryption algorithm. It is used as a common measure to analyze the quality of encryption methods in terms of how it maximizes the deviation between original and the cipher image . Histogram deviation can be measured using Eq. 3.
Histogram analysis results
Deviation from ideality
5.2.3 Irregular deviation
5.2.4 Deviation from ideality
Table 4 shows the values of irregular deviation, histogram deviation and deviation from ideality obtained for various medical images when encrypted. The results show that the encrypted image has a high histogram deviation from the original image and negligible irregular deviation and minimum deviation from ideally encrypted image.
5.3 Correlation analysis
5.3.1 Correlation distribution
From the obtained results, it can be concluded that the proposed encryption scheme minimizes the correlation among pixels in the plain image.
5.3.2 Correlation coefficient analysis
Correlation between plain and cipher images
5.3.3 Pixel correlation
Correlation among adjacent pixels
5.4 Analysis of differential attack
Analysis of differential attack is testing of how the change in information of the input affects the output. Differential attack is applicable primarily to block ciphers working on binary sequences , and it becomes a common attack to be considered in cipher design. The widely used methods of differential attack in image processing are analysis of NPCR and UACI metrics.
5.4.1 NPCR analysis
NPCR and UACI results
5.4.2 UACI analysis
5.5 Key space analysis
Results when image is decrypted with key differ in one base
Correlation with the plain image
Correlation of horizontal pixels
Correlation of vertical pixels
Correlation of diagonal pixels
5.6 Quality analysis
5.6.1 Peak signal to noise ratio(PSNR)
5.6.2 Information entropy and SSIM
Values obtained for PSNR, MSE, image fidelity, structural content (SC), information entropy and SSIM
Entropy of plain image
Entropy of cipher image
The loss of structural information due to encryption can be estimated by calculating SSIM(structural similarity in images)  of the cipher image. SSIM is used to compare the contrast luminance and structure of two images. The obtained entropy shows the randomness generated in the image after encryption. Also, the SSIM obtained ensures the dissimilarity between plain nd encrypted images.
5.6.3 Mean square error(MSE)
High value of MSE in Table 9 represents robustness to differential attack and thereby ensuring the quality of encryption scheme.
5.6.4 Image fidelity(IF) and structural content(SC)
The proposed method minimum information fidelity and structural content, from which the quality of the scheme is guaranteed.
The obtained values of PSNR, MSE, IF, SC, entropy and SSIM shown in Table 9 assure quality of the proposed method.
5.7 Mean value analysis
Comparison with correlation coefficients
Comparison with existing methods
The paper proposes a secure encryption method based on DNA cryptography and two-dimensional Zaslavski map for various medical images. There are two phases: permutation and diffusion. A detailed performance analysis is done to test, analyze and ensure the strength, security and quality of the method. Randomness of the pseudorandom sequence which is used in permutation is tested by various randomness tests and autocorrelation analysis. Visual analysis is done by various histogram analysis methods. High value of NPCR and UACI shows that the proposed encryption scheme is robust against differential attack. Apart from this, the quality of the encryption method is tested. The experimental results show that a slight variation in the key yields a highly uncorrelated image compared to the plain image. The results are found to be authentic, and hence, the method is well suited for medical image encryption.
7 Future scope
The method can be enhanced by encrypting the extracted features of the image other than the whole image. This can reduce time required for execution.
Compliance with ethical standards
Conflict of interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.
- 1.Zhou Y, Panetta K (2009) A lossless encryption method for medical images using edge maps. 31st annual ınternational conference of the IEEE EMBS, Minneapolis, Minnesota, USA, 2–6 September 2009Google Scholar
- 3.Mondal B, Mandal T (2017) A light weight secure image encryption scheme based on chaos & DNA computing. J King Saud Univ Comput Inf Sci 29:499–504Google Scholar
- 8.Chirakkarottu S, Mathew S (2018) A comparative analysis of image encryption techniques using chaotic maps and DNA cryptography. J Adv Res Dyn Control Syst 10:301–314 15-Special Issue Google Scholar
- 9.Stoyanov B, Kordov K (2014) Novel Zaslavsky map based pseudorandom bit generation scheme. Appl Math Sci 8(178):8883–8887Google Scholar
- i.Ramadan N, Ahmed HEH, Elkhamy SE, El-Samie FEA (2016) Chaos-based image encryption using an ımproved quadratic chaotic map. Am J Signal Process 6(1):1–13Google Scholar
- 14.Stoyanov B, Kordov K (2014) Novel zaslavsky map based pseudorandom bit generation scheme. Appl Math Sci 8(178):8883–8887Google Scholar
- 15.Ahmad J, Ahmed F (2010) Efficiency analysis and security evaluation of image encryption schemes. IJVIPNS-IJENS, pp 18–31Google Scholar
- 16.Hor A, Ziou D (2010) Image quality metrics: PSNR versus SSIM. 2010 ınternational conference on pattern recognitionGoogle Scholar
- 17.Rajput Y, Gulve AK (2014) A comparative performance analysis of an image encryption technique using extended Hill Cipher. Int Journal of Comput Appl 95(4):17–20Google Scholar
- 18.Kumar R, Rattan M (2012) Analysis of various quality metrics for medical image processing. Int J Adv Res Comput Sci Softw Eng 2(11):137–144Google Scholar