Elsevier

Applied Soft Computing

Volume 51, February 2017, Pages 168-179
Applied Soft Computing

Protection of medical images and patient related information in healthcare: Using an intelligent and reversible watermarking technique

https://doi.org/10.1016/j.asoc.2016.11.044Get rights and content

Highlights

  • An intelligent reversible watermarking technique for the protection of medical images.

  • The concept of companding function is exploited for reducing embedding distortion.

  • Integer Wavelet Transform is used as an embedding domain for achieving reversibility.

  • Histogram processing is employed to avoid underflow/overflow.

  • GP is used to evolve models that not only make an optimal tradeoff between imperceptibility and capacity of watermark.

Abstract

This work presents an intelligent technique based on reversible watermarking for protecting patient and medical related information. In the proposed technique ‘IRW-Med’, the concept of companding function is exploited for reducing embedding distortion, while Integer Wavelet Transform (IWT) is used as an embedding domain for achieving reversibility. Histogram processing is employed to avoid underflow/overflow. In addition, the learning capabilities of Genetic Programming (GP) are exploited for intelligent wavelet coefficient selection. In this context, GP is used to evolve models that not only make an optimal tradeoff between imperceptibility and capacity of the watermark, but also exploit the wavelet coefficient hidden dependencies and information related to the type of sub band. The novelty of the proposed IRW-Med technique lies in its ability to generate a model that can find optimal wavelet coefficients for embedding, and also acts as a companding factor for watermark embedding. The proposed IRW-Med is thus able to embed watermark with low distortion, take out the hidden information, and also recovers the original image. The proposed IRW-Med technique is effective with respect to capacity and imperceptibility and effectiveness is demonstrated through experimental comparisons with existing techniques using standard images as well as a publically available medical image dataset.

Introduction

The advancement in communication technologies has provided new ways of accessing and transferring the medical information. The widespread use of information and communication techniques allows one to manipulate the original contents. Therefore, to protect the patient privacy and to ensure diagnostic accuracy, an effective medical image authentication mechanism is required without losing the semantics of the original content [33]. In this regard, digital watermarking has been advocated by many researchers as one of the most promising techniques to provide security, reliability, and authenticity of medical information [5].

Watermarking is considered as the practice of imperceptibility [16], changing a work to embed some information related to the work. Temper detection [22], copyright control [4], owner identification [6], broadcast monitoring [11], etc., are some of the interesting examples of watermarking applications. Generally, watermarking is categorized as Fragile, Robust, and Semi-fragile watermarking [10], [20], [7]. In fragile watermarking, alteration in image results in the destruction of watermark and robust watermarking aims to resist attacks [34], [30]. Whereas in semi fragile watermarking, a watermark is supposed to resist unintentional attacks but get destroyed in case of intentional attacks.

The medical image watermarking is quite challenging and some essential constraints must be considered during the watermarking process. Embedding information in the host image causes distortion, which may be quite detrimental for medicalor military applications. A minute distortion in the medical image due to watermark embedding may turn it impracticable for the physicians. Therefore, there is a strong need to develop a watermarking technique that is not only able to embed the watermark but also able to restore the original content of the image after the watermark extraction. Reversible watermarking has the capability to restore the image back to the exact state, and thus fulfills the basic requirement [9]. An efficient reversible watermarking should be capable of embedding more information with fewer perceptual distortions as well as restoring the original image content. However, watermark capacity and imperceptibility are two contradicting properties, and therefore, it becomes a challenging task to make an optimal tradeoff between them for a given image and the intended application.

In medical image watermarking, one conventional way of protecting useful information in medical images is by defining a region of interest (ROI). The watermark is only embedded in ROI thus protecting the useful information from distortion [32]. However, in most scenarios the whole medical image has to be considered as ROI and embedding of information using ROI based approaches may not be acceptable.

This paper presents a block-based watermarking approach, in which the proposed IRW-Med utilizes the concept of companding for watermark embedding. Moreover, Genetic programming (GP) based intelligent coefficient selection is performed in integer wavelet domain, to exploit the inter-dependencies of wavelet coefficients.

The main leverage of this paper is in developing an intelligent embedding model for the given image, which has the following capabilities:

  • a)

    To make a suitable trade-off between capacity and imperceptibility of the watermark. Moreover to exploit the dependencies of wavelet coefficients and information regarding the type of sub bands.

  • b)

    To develop a model that not only helps in the selection of suitable coefficient for watermark embedding but also acts as the companding factor during embedding of the watermark.

In the proposed IRW-Med, block-based embedding helps not only in evolving mathematical expressions that select coefficients for companding through GP, but also acts as a threshold for companding. It is shown that for a given image (depending upon its frequency content), the learning ability of the GP makes it possible to find a suitable tradeoff between the watermark payload and imperceptibility. The optimal tradeoff between watermark imperceptibility and capacity is required in medical image watermarking for avoiding any misdiagnosis. The remaining portion of the paper is classified in the following way. Details regarding the proposed IRW-Med are described in Section 2. Results and performance analysis are discussed in Section 3. GP based implementation details are explained in Section 4 and at the end the discussion related to conclusion is explained in Section 5.

Section snippets

Related work

For decades researchers have put great efforts in the domain of reversible watermarking. Fridrich [13] proposed a lossless bit plane compression where extra space that is retrieved is used for the embedding of both the watermark and bookkeeping data. On the other hand, Alattar [1] proposed difference expansion to quads as a data hiding approach and achieved high capacity. Xuan et al. [26] proposed lossless approach related to data hiding based on Interger Wavelet Transform (IWT) and threshold

Proposed GP-IRW-Med approach

Like conventional watermarking, the proposed IRW-Med technique has two key phases; Embedding phase and Extraction phase. Proposed IRW-Med employs GP to evolve a mathematical function that selects the coefficients for companding. The general structural design of our proposed approach is based on the training phase (development of selection model of wavelet coefficients) and testing phase as shown in Fig. 1.

Results and discussion

The proposed IRW-Med approach is tested on various gray scale standard images such as Lena, Baboon, Barbara, and Gold hill as well as on standard medical images; the size of each image is of size 512 × 512. Matlab based GPLab toolbox has been utilized in order to develop a suitable GP expression. For Lena image best expression is presented in prefix notation as follows.

Conclusions

In this paper, reversible watermarking technique GP IRW-Med based on GP for protecting both the patient related information and associated medical image is proposed. Embedded watermark is based on the medical related information e.g., we have embedded EPR (Electronic patient record) as watermark, however embedding procedure is generic in nature.The capability of GP to learn is used to find suitable tradeoff between the imperceptibility and payload. GP exploits the hidden characteristics of

Acknowledgment

This work is supported by the Higher Education Commission of Pakistan under NRPU Research Grant Nos. 20-1624/R&D/10/4603 and ICTRDF/TR&D/2012/62.

References (37)

  • D. Coltuc

    Improved embedding for prediction-based

    IEEE Trans. Inf. Forensics Secur.

    (2011)
  • I. Cox et al.

    Digital Watermarking

    (2001)
  • I.J. Cox et al.

    Secure spread spectrum watermarking for images, audio and video

  • G. Depovere, T. Kalker, J. Haitsma, M. Maes, L. De Strycker, P. Termont, et al. The VIVA project: digital watermarking...
  • I.-C. Dragoi et al.

    Local-prediction-based difference expansion reversible watermarking

    IEEE Trans. Image Process.

    (2014)
  • J. Fridrich et al.

    Invertible authentication

  • X. Gao et al.

    Lossless data embedding using generalized statistical quantity histogram

    IEEE Trans. Circuits Syst. Video Technol.

    (2011)
  • R.C. Gonzalez et al.

    Digital Image Processing

    (2007)
  • Cited by (60)

    • From classical to soft computing based watermarking techniques: A comprehensive review

      2023, Future Generation Computer Systems
      Citation Excerpt :

      These are more resilient in comparison to the spatial domain. Some popular techniques of frequency domain are DFT [9,62], DCT [63,64], DWT [65], LWT [66,67], RDWT [13,68], and SVD as depicted in Fig. 4. Algorithm 2 discusses the general procedure of embedding and extracting a watermark in the frequency domain.

    • Reversible image authentication scheme with blind content reconstruction based on compressed sensing

      2022, Engineering Science and Technology, an International Journal
      Citation Excerpt :

      Although several data hiding methods have been designed to provide high embedding capacity, most of them are not reversible. Reversible data hiding methods based on integer wavelet transform (IWT) fulfill these requirements [7,9,24]. IWT provides an integer-to-integer transformation, which is essential to preserve reversibility as round-off errors are avoided.

    • A novel zero-watermarking scheme with enhanced distinguishability and robustness for volumetric medical imaging

      2021, Signal Processing: Image Communication
      Citation Excerpt :

      Therefore, the motivation of our work is to design a novel watermarking scheme to satisfy these additional requirements. Three categories of digital watermarking schemes are suitable for authenticity and copyright protection of medical images, which are ROI lossless watermarking [10–18]; reversible watermarking [19–29]; and zero-watermarking [30–40]. ROI lossless watermarking schemes embed watermark into medical images while keeping their ROI regions distortion-free to minimize the bias on medical diagnosis.

    View all citing articles on Scopus
    View full text