High capacity data hiding schemes for medical images based on difference expansion

https://doi.org/10.1016/j.jss.2010.08.055Get rights and content

Abstract

Since the difference expansion (DE) technique was proposed, many researchers tried to, improve its performance in terms of hiding capacity and visual quality. In this paper, a new scheme, based on DE is proposed in order to increase the hiding capacity for medical images. One of the characteristics of medical images, among the other types of images, is the large smooth regions. Taking advantage of this characteristic, our scheme divides the image into two regions; smooth region and non-smooth region. For the smooth region, a high embedding capacity scheme is applied, while the original DE method is applied to the non-smooth region. Sixteen DICOM images of different modalities were used for testing the proposed schemes. The results showed that the proposed scheme has higher hiding capacity compared to the original schemes.

Introduction

Medical images are a very important part of patient's records and information, which are stored in databases of hospitals and may be exchanged between hospitals and health centers. Among these data, both patient information and medical images need to be properly organized, so as to avoid mishandling and loss of data (Nayak et al., 2008). In order to achieve an efficient utilization of the communication channel bandwidth and storage, data size must be reduced. Separate transmission of the image and data requires more bandwidth in transmission and more memory space during storage. Interleaving one form of data, such as 1D signal or text files, over digital images can combine the advantages of data security with efficient memory utilization (Nayak et al., 2004).

Many data hiding techniques are used for interleaving patient information with medical images. Moreover, these data hiding techniques can also be used for authentication and tamper detection to judge images integrity and fidelity (Al-Qershi and Ee, 2009). In order to achieve that, much data must be concealed into the image besides the patient's data. Thus, the capacity of the hiding must be high enough to accommodate the payload. On the other hand, reversibility is one of the most important requirements for medical images, which must be kept intact to avoid any misdiagnoses (Kamstra and Heijmans, 2005).

One of the outstanding reversible data hiding schemes is the difference expansion (DE), which was introduced by Tian (2003). The main advantage of DE, among the other reversible schemes, is the high embedding capacity which is achieved by this method (Thodi and Rodríguez, 2007). DE is based on modifying the difference between a pair of pixel values while keeping the average of them unchanged. His technique divides the image into pairs of pixels, then embeds one bit of information into each pair. The technique received more attention over the years, because of its high efficiency and simplicity. Since Tian introduced the difference expansion technique, the influence of his work can be easily found in many later works. The extensions of his method by other researchers have yielded many good and more sophisticated algorithms (Hu et al., 2006).

The early improvement of Tian's technique was proposed by Alattar. He aimed at increasing the hiding capacity by embedding data in triplets (Alattar, 2003), then in quads (Alattar, 2004a), and later in vectors (Alattar, 2004b). The proposed schemes can achieve high embedding capacity without losing the simplicity of the original DE technique. Another simple DE-based data hiding scheme is the one proposed by Chiang et al. (2008). Instead of embedding data in pairs or quads, the data are embedded into smooth blocks of 4 × 4 pixels in order to increase the hiding capacity.

The aims of improving the original DE proposed by researchers are twofold: first is to make the embedding capacity as high as possible, second is to make the visible distortion as low as possible. To achieve high embedding capacity, the reviewed schemes adopted three different approaches: (i) simplifying the location map in order to increase its compressibility, (ii) embedding payload without location map, and (iii) expanding differences more than once which allows more data to be embedded. Meanwhile, the visual quality may be enhanced by: (i) using a predefined threshold T, (ii) selecting smooth areas to embed data, and (iii) using sophisticated classification functions. However, there is a trade off between distortions and embedding capacity. If distortion is minimized, only a few data can be embedded. On the other hand, if the embedding capacity is increased, it results in low visible quality.

Although, the reviewed schemes can achieve the two aims mentioned above in a relative way, each scheme has its own drawbacks. The drawbacks can by summarized as follows:

  • 1.

    Some of the schemes did not display good visual quality for all embedding frequencies (Kamstra and Heijmans, 2005, Thodi and Rodríguez, 2007, Alattar, 2003, Alattar, 2004a, Alattar, 2004b, Weng et al., 2007, Lee et al., 2008).

  • 2.

    Some schemes need side information to initiate the extracting phase, while these data are not embedded into the host image (Lee et al., 2008, Lin et al., 2008, Chang et al., 2008).

  • 3.

    Some of the schemes show good performance only with smooth images, which makes them suitable for specific types of images (Chiang et al., 2008).

  • 4.

    Some of the schemes are complicated (Weng et al., 2007, Lee et al., 2008, Kim et al., 2008).

  • 5.

    All of the schemes are time consuming, because they scan the host image more than once.

To overcome those drawbacks and improve the performance of DE-based schemes, a combination of two or more schemes may be used in order to get the advantages of them together.

In this paper, we aim at obtaining a reversible data hiding scheme that can achieve high embedding capacity for medical images. We propose two combined data hiding schemes based on DE for medical images. In the first scheme, we combine Chiang's scheme with Alattars’ scheme, and in the second one we combine the Chiang's scheme with Tian's scheme. Each one of the three schemes has the characteristics of simplicity and high embedding capacity, and by combining them, we can maximize the hiding capacity. Of course there are some other DE-based schemes which can be found in the literature, but most of them lack simplicity (Kamstra and Heijmans, 2005, Weng et al., 2007, Kim et al., 2008).

This paper is organized as follows. The three above-mentioned schemes are reviewed in Section 2. In Section 3, the two reversible schemes are proposed. The experimental results are displayed in Section 4. Finally, the results are discussed and concluded in Section 5.

Section snippets

Overview

In this section, the three DE-based data hiding schemes, which are used in our combined schemes, are reviewed.

The proposed schemes

For a smooth block of 4 × 4 pixels, the maximum number of bits that can be embedded using schemes of Tian, Alattar, and Chiang separately is 8, 12, and 14, respectively. This means that Chiang's scheme is the most appropriate scheme for smooth blocks. However, for non-smooth blocks, the number of data bits that can be embedded is zero. Thus, our scheme combines Chiang's scheme with Tian's and Alattar's schemes in order to get the advantages of them together. Two schemes are proposed in this

Experimental results

Digital Imaging and Communications in Medicine (DICOM) format was used for testing the proposed schemes. The used images were of different modalities and sizes as shown in Fig. 2. A random bit stream was generated as payload. The two proposed schemes, Tian's technique, Alattar's, and Chiang's schemes were used to embed the payload into test images separately. For Alattar's and the 2nd proposed scheme, a threshold of T = 30 is used with US images, while T = 300 was used for the rest. That is because

Discussion

In this paper, two reversible schemes were proposed for data hiding in medical images. The first scheme combined Tian's technique with Chiang's scheme, and the second scheme combined Tian's technique with Alattar's scheme. The results obtained asserted the fact that Chiang's scheme is image-dependent. This can be noticed from the wide range of available space for the 16 images used for testing, 0.001 bpp for image X-Ray3 image in comparison to 0.485 bpp for image US1 as shown in Table 1. The

Conclusion

Many data hiding schemes were proposed in the last few years, but few of them were proposed originally for medical images. Those schemes, which may work perfectly with general images, may not show the same performance with medical images.

In this paper, two reversible data hiding schemes based on DE are proposed for medical images. Both of the two proposed schemes are simple and easy to implement, but image-dependent. Nevertheless, the first scheme displayed high hiding capacity while keeping

Acknowledgment

This work is supported by Ministry of Science, Technology and Innovation, Malaysia through eScienceFund grant 01-01-05-SF0114 and Ministry of Higher Education, Malaysia through Fundamental Research Grant Scheme (203/PELECT/6071135). The authors wish to thank Assoc. Prof. Dr. Mohd Ezane Aziz, School of Medical Sciences, Universiti Sains Malaysia for providing the medical images.

Osamah M. Al-Qershi completed his Bachelor of Science (B.Sc) in computer control engineering from University of Technology Baghdad, Iraq in 1998. He is currently a master student in the School of Electrical & Electronic Engineering at Universiti Sains Malaysia (University of Science, Malaysia). His research interest is in the area of digital image watermarking and authentication.

References (17)

  • C.-C. Lee et al.

    Adaptive lossless steganographic scheme with centralized difference expansion

    Pattern Recognition

    (2008)
  • A.M. Alattar

    Reversible watermark using difference expansion of triplets

  • A.M. Alattar

    Reversible watermark using difference expansion of quads

  • A.M. Alattar

    Reversible watermark using the difference expansion of a generalized integer transform

    IEEE Transactions on Image Processing

    (2004)
  • O.M. Al-Qershi et al.

    Authentication and data hiding using a reversible ROI-based watermarking scheme for DICOM images

  • Chang, Z., Xu, J., Kou, W., 2008. “Reversible Watermarking Schemes Using Spatial Quad-Based Difference Expansion,” in:...
  • K.-H. Chiang et al.

    Tamper detection and restoring system for medical images using wavelet-based reversible data embedding

    Journal of Digital Imaging

    (2008)
  • Y. Hu et al.

    Analysis and comparison of typical reversible watermarking methods

    Lecture Notes in Computer Science

    (2006)
There are more references available in the full text version of this article.

Cited by (82)

  • High-capacity reversible data hiding in encrypted images based on pixel-value-ordering and histogram shifting

    2023, Expert Systems with Applications
    Citation Excerpt :

    Reversible data hiding (RDH) refers to the technology that can avoid permanent damage of the cover image while some secret data is embedded into the cover image (He et al., 2018; Ou et al., 2016; Shi et al., 2016). Up to now, a lot of RDH methods have been proposed and they can be classified into three categories (Shi et al., 2016): difference expansion based methods (Al-Qershi & Khoo, 2011; He et al., 2018; Ou et al., 2016), histogram shifting (HS) based methods (He et al., 2016; Tai et al., 2009) and lossless compression based methods (Lin et al., 2015; Zhang et al., 2013). However, image transmission is vulnerable to security threats such as hacker hijacking, leaks and eavesdropping.

  • Information hiding scheme for digital images using difference expansion and modulus function

    2019, Journal of King Saud University - Computer and Information Sciences
View all citing articles on Scopus

Osamah M. Al-Qershi completed his Bachelor of Science (B.Sc) in computer control engineering from University of Technology Baghdad, Iraq in 1998. He is currently a master student in the School of Electrical & Electronic Engineering at Universiti Sains Malaysia (University of Science, Malaysia). His research interest is in the area of digital image watermarking and authentication.

B.E. Khoo received the B.Tech degree in Quality Control and Instrumentation from Universiti Sains Malaysia in 1993, and PhD. degree in Electrical Engineering from University of Wales, Swansea in 1998. She is currently a senior lecturer with School of Electrical and Electronic Engineering, Universiti Sains Malaysia. Her current research interests include digital watermarking, computer vision and multimedia forensics.

View full text