Research Of Medical Image Processing Based On Wavelet Analysis

With the development of technology, medical imaging technology plays an increasingly important role in clinical diagnosis and treatment. Medical images, such as CT, Ultrasonic B-scanner and MRI, can provide relatively objective data for doctors, discover diseases on early stage and are auxiliary means for doctors to cure diseases. However, medical image itself has many flaws. Sometimes the border is unclear or appears darkly. In addition, the image quality will be influenced by the equipment's photography mechanism, the interference during transiting or display unit during imaging process, which makes it hard to get an accurate judgment of the image and brings some difficulties for the doctor to diagnose and treat diseases. In order to improve the readability of the medical images and diagnose patient's condition effectively, medical images will be processed.Traditional image processing methods are easy to magnify the noise, and lose the edge or detail, which are not suitable for processing medical images. Wavelet analysis is the new frequency analysis tool after the Fourier analysis. It has functions as follow: time-frequency localization, multi-resolution analysis and signal-noise to separate and gains distinct advantage in handling medical images. This dissertation mainly studies enhancement and de-noising algorithm of medical images based on wavelet transform and mainly does work as follows:First, several major modules of the PACS are introduced. It mainly exhibits basic theories of wavelet transform and image decomposition and reconstruction, which provides the theoretic basis for the subsequent image processing.Second, we analyze modulus maximum de-noising, relevance de-noising and threshold de-noising methods. Then we propose a new threshold function based on hard threshold and soft threshold functions, and prove the new threshold function performs better than hard threshold and soft threshold functions through experiments. After that, we make a further analysis of NeighShrink threshold de-noising. Since NeighShrink method takes neighborhood wavelet coefficients into consideration while dealing with wavelet coefficients, this reduces the loss of important details, keeps the edge and details of the image, and overcomes the shortcomings of soft threshold and hard threshold de-noising methods. However, this method will minimize all the wavelet coefficients while de-noising, which diminishes the edge or details of images. To solve this problem, an improved method is proposed, which can enhance the edge and detail information of the image while de-noising. Simulation results show that the improved method performs better than wavelet compulsory de-noising, soft threshold de-noising, hard threshold de-noising and NeighShrink threshold de-noising method.Finally, we study image enhancement method based on wavelet transform and discuss the algorithm of high frequency sub-bands of wavelet in detail. Then a wavelet enhancement algorithm based on a new threshold function is proposed, and the image after wavelet enhancement will be processed by Laplacian enhancement. Experiments show that Laplacian enhancement method of wavelet not only has prominent edge and detail information, but also prevents the magnification of noise, whose effect is better than the single method.