Medical Image Encryption Using Edge Maps

Introduction

This paper presents a medical image encryption algorithm using edge maps derived from a source image. The algorithm is composed by three parts: bit-plane decomposition, generator of random sequence, and permutation. It offers users the following flexibilities: (1) any type of images can be used as the source image; (2) different edge maps can be generated by various edge detectors and thresholds; (3) selection of appropriate bit-plane decomposition method is flexible; (4) many permutation methods can be cascaded with the proposed algorithm. A significantly large key space and strong key sensitive are possessed by the proposed algorithm to protect different types of medical images. Furthermore, it has a wider applicability than other methods for fuzzy edge maps. The histograms of the cipher-images are approximately flat even with blurry edge maps, which verify EMMIE possessing a strong robustness for fuzzy edge maps by the simulation results. The analysis of security demonstrates EMMIE is a secure algorithm. Source image, edge detector, and the parameters of scrambling method compose the keys of EMMIE and their combinations are significantly large to defend the Bruce-force attack. To further evaluates the security level of EMMIE, as is shown in the comparisons with other state-of-the-art methods, EMMIE possesses a higher pixel correlation, stronger key sensitivity and error robustness, and a better performance against differential attack with less time cost.

Main results

weijiacao5

EMMIE encryption results of the CT image: (a) The edge map extracted from Fig. 3 (q) (Sobel detector); (b) the cipher-image; (c) the reconstructed CT image; (d) the histogram of the original image shown in the Fig. 3(p); (e) the histogram of (b); (f) the histogram of the difference between the original image and (c).

weijiacao6

Correlation coefficients of the original image shown in the Fig. 9(a) and cipher-images using different encryption algorithms.

weijiacao7

The encryption time of the schemes of Zhu, Zhou, DecomCrypt and EMMIE with the increasing size of the original image in Fig. 3(l). The green, red, blue-green and blue lines represent the methods of Zhu, Zhou, DecomCrypt and EMMIE, respectively.

weijiacao8

NPCR and UACI results of different encryption algorithms.