Abstract
Digitalization of media has exploded in recent years. It has resulted in the rise of private data hacking, which has been increased by the growth of the data exchange system, i.e., the Internet, as well as the simple access to storage media. New approaches, such as watermarking in (C. Iwendi et al., 2020; D. Datta et al, 2021; Randhir Kumar et al,2021), are being used to combat these hackers. The application of image watermarking technologies to medical images, as proposed in (Nazari, M., et al, 2021; Thanki, R, 2021; Manoj K., 2020), is the focus of this research. In this chapter, we propose a new robust blind crypto-watermarking solution for medical imaging or DICOM file (Digital Imaging and Communications in Medicine) security based on masking (or hide) electronic patient information (patient name, patient ID, patient age...) in its medical imaging, then, erases them from the tag of the DICOM. Before being included into medical imaging, DICOM patient information, or EPR, is encrypted using a modified AES (Advanced Encryption Standard) encryption technique. The image is broken into 8x8 pixel chunks. In each block, we use the 2D-LWT (Lifting wavelet transform), 2D-DCT (discrete cosine transforms), and SVD (singular value decomposition) to insert one bit of the encrypted watermark into the hybrid transform domain. Various attacks, such as noise, filtering, scaling, and compression, are used to test the method. According to the obtained results the watermark (EPR) is imperceptible in the imaging, and the suggested technique has passed the attacks test with success.
TopIntroduction
The use of telemedicine (Kumar et al., 2017; Lakshmanna et al., 2016) has helped patients distancing in hospitals during the Covid19 (Coronavirus Disease 2019) pandemic and its danger on persons with chronic conditions, as described in (Rajput et al., 2021). Medical data, diagnostics, medical images, and reports are sent between health entities using information and communications technology. The necessity to safe these data has become important as a result of this rapid change. Digital watermarking (Boussif et al., 2017, 2020; Saeid et al., 2022; Gutub, 2022; Hassan et al., 2021; Adnan, 2010), cryptography (Mohamed et al., 2019, 2020; Koppu et al., 2018, 2020; Adnan, 2020; Adnan et al., 2021), and crypto-watermarking (Hureib et al., 2020 ; Noorah et al., 2018) are now widely employed in the network environment to protect intellectual property, and they have become the principal applications for protecting and securing medical images. Ensuring the security of DICOM data must respect the imperceptibility of electronic patient information when hidden in medical imaging. The watermark must be extremely resistant to many forms of attacks. Several approaches for watermarking of medical images have been suggested in the State-Of-The-Art, such as (Mettripun, 2016), where the author developed a robust medical imaging watermarking approach based on DWT (Discrete Wavelet Transform) for patient identification. The context is identical to this chapter. However, he has only employed the DWT transform which making the system vulnerable to some attacks such as geometry transformation. By combining the DCT with the DWT, M. Jamali et al. (2016) presented a robust watermarking approach in Non-ROI (Region of Interest) of the medical images. The insertion was precisely in a specific area of the imaging which resulting a small insertion capacity. B. Kima et al. (2003) was suggested a digital image watermarking scheme that is particularly resistant to geometrical attacks. However, because filtering and compression are widely utilized in DICOM medical imaging, the resistance against these attacks is required. Y. Zolotavkin et al. (2014) suggested a novel QIM (Quantization Index Modulation)-based watermarking approach that is robust to gain attack. However, because medical images are sensitive, imaging quantization is not suitable for DICOM images. To ensure security in Telemedicine, Z. Ali et al. (2018) suggested a unique watermarking scheme based on the Hurst Exponent. C. Manuel et al. (2015) presented a DFT (Discrete Fourier Transform)-based watermarking system for the protecting of medical imaging in hospitals. Rayachoti Eswaraiah et al. (2015) presented a robust image watermarking approach for identifying tampers inside ROI and retrieving the original ROI. Due to a combination of FDCuT (Fast Discrete Curvelet Transform) and DCT, Rohit Thanki et al. (2017) presented a novel watermarking approach. In invariant DWT, Y. Gangadhar et al. (2018) introduced a watermarking strategy for safe medical images. Priya Selvam et al. (2017) suggested a reversible watermarking solution for medical imaging security in telemedicine applications that combines signal processing transformations. A high-capacity watermarking approach for DICOM images has been proposed by Frank Y. Shih et al. (2016).
Proposed watermarking algorithms need improvement by increasing the robustness of the watermark since transmission and storing images in public networks and clouds are exposed to many attacks like noise and compression, respectively, that can remove the watermark. For this issue, in this paper, we propose a novel robust blind crypto-watermarking system for the security of DICOM images in telemedicine. As shown in Figure 1, we encrypt the watermark (DICOM patient information) before inserting it in the imaging. We propose a watermarking algorithm combining the LWT-DCT-SVD transforms to be robust. The AES encrypted data are inserted block by block in the cover image. The objectives can be summarized as follows: 1) Hide the information of patients in the related imaging. 2) Attach imaging with its relative patient. A short version of this chapter that use AES with 128 bits has been presented at an international conference (Boussif, 2021). In this chapter, we use a modified AES to be more suitable with the watermarking process.
There are many approaches in the literature that have propose a watermarking system for DICOM images, which hide patient identities in the imaging, like in our chapter. However, the system will be more efficient if the watermarking is more robust.