Article Preview
Top2. Requirements Storage Needs Of The Epr In Medical Images
Most watermarking techniques counting on modify and hence distort the host image in order to embed the given data. In many applications, loss of image fidelity is not prohibitive as long as the original and modified images are perceptually equivalent. In medical applications, where the need for authentication is often paramount, there are typically stringent constraints on data fidelity that prohibit any distortion during the watermarking process. In certain cases, artifacts in a patient’s diagnostic image may cause errors in diagnosis and treatment with possible life threatening consequences (Zain et al., 2009).
A well-known medical image standard DICOM (Digital Image Communication in Medicine) can be used to insert EPR data in to an image, but this method has a risk of data loss in case of anonymization process. Even Munch et al. (2004) proposed a web based method to integrate the data and images but the best way of integration of EPR and medical images is the hiding EPR in images using watermarking techniques.
The frequently used approach in spatial domain is Least Significant Bit (LSB) insertion. Since the least significant bit-plane does not contain visually significant information, it can be easily replaced by a large amount of watermark bits. But this technique doesn’t satisfy medical requirements in terms of data integrity. In order to use it, we propose to compress the original LSB bit-plan using lossless technique that offers high level compression.
Run Length Encoding (RLE) is a well known lossless compression algorithm; it only offers decent compression ratios with files that contain lots of repetitive data. Several other compression techniques exist in the literature; we opt for Huffman coding algorithm (Hu & Chang, 2000) which offers a reasonable compression rate and it is adapted to binary data exiting in the LSB bit-plan.
To ensure high security, data encryption is unavoidable. But nowadays, online diagnosis is widely used, which means we should find an efficient and speedy encryption algorithm.