Document Type: Original Paper
M. Sc. in Biomedical Engineering, Tehran University of Medical Sciences, Tehran, Iran.
Assistant Professor, Physics and Biomedical Engineering Dept., Tehran University of Medical Sciences, Tehran, Iran. Research Center for Science & Technology in Medicine, Imam Khomeini Hospital, Tehran, Iran.
Reader in UWB communications, King’s college London, University of London, Center of Telecommunications Research, London, UK.
Lecturer in UWB Communications, King's College London, University of London, Center of Telecommunications Research, London, UK.
Introduction:In this study, ECG signals have been embedded into medical images to create a novel blind watermarking method. The embedding is done when the original image is compressed using the EZW algorithm. The extraction process is performed at the decompression time of the watermarked image.
Materials and Methods: The multi-resolution watermarking with a secret key algorithm developed in this work is blind because during the EZW decoding only the secret key which includes the header information is used to extract the embedded information. There are two processes added to EZW coder including insertion process and extraction process.
Results: Several grey scale MRI and CT images of size 256 x 256 and 512 x 512 pixels have been watermarked by embedding five different watermarks of size 512, 1024, 2048, 4096, 8192 bytes. The used images were divided into three categories such as MRI, CT and Mammography. Also they are in the size of 256×256 and 512×512. The ECG signal is one kind of normal ECG signal that has been utilized for experiments.
In this experiment, the PSNR obtained between the original image and the compressed watermarked was found to be greater than 35 dB for all medical images from different modalities. This finding was also true for the PSNR between the original mark signal and its reconstruction. The β measured for the watermarked image was greater than 0.7 and for most of the images it was greater than 0.9.
Discussion and Conclusion: The achieved PSR and β prove the imperceptibility of the method in addition to preserving the edges of the host image. The proposed method is able to utilize about 15% of the host image to embed mark signal that is suitable for most watermarking methods. The proposed method of selecting the insertion sites ensures three watermarking constraints: the Imperceptibility and Robustness, High percentage of watermark and Blindness.