Research On Preprocessing Algorithm Of Medical Ultrasound Image Based On ICA And Wavelet Transform

The ultrasonic diagnose is one of the irreplaceable imaging diagnostic techniques in the modern clinical medicine, due to its merits of non-destructive, low cost, no radiation and real-time imaging. Because of the characteristic of the coherent imaging process, which produce a lot of speckle noise, ultrasound images suffer from a lot of consequences, such as low signal-to-noise ratio and poor image quality, even cover and reduce the image in some detail. It brings unfavorable effects into the image feature extraction and recognition, moreover the quantitative analysis and diagnosis. So, suppress the noise, enhance the image details and improve image quality are the important part of pretreatment in ultrasound image analysis and recognition.The medical ultrasound image preprocessing methods are reviewed. And we focus on the medical ultrasound image denoising and enhancement methods. Two improved algorithms are proposed addressing the existing problems:Compared with the traditional denoising methods, ICA algorithm has obvious advantages in the separation of signal and noise. In practical problems, however, the dimension of the source signal dimension and the observed signal are not equal, thereby, the algorithm is limited application. To solve this problem, the paper proposes a novel method named ICA algorithm based on virtual channels. The algorithm introduces the concept of virtual channels to increase the dimension of the observation signals. Then it analyzes the principle of virtual channels used in ICA, and discusses how to construct a virtual channel. Finally, the denoising performance of the algorithm is validated; the advantages and advantages are analyzed. It is the foundation of future study.Whereas, if the ICA algorithm which based on virtual channels is worked on the medical ultrasound image denoising. The problem of how to obtain high matching virtual channels without preknowledge should be solved firstly. For this reason, this paper proposes an medical ultrasound image denoising algorithm based on adaptive virtual channel. Considering with the structure problems of the virtual channel, it uses DT-CWT to decompose the medical ultrasound images, constructing the virtual channels adaptively.Then it discusses the denoising advantages of combining DT-CWT with the ICA. Finally, comparing with the classical medical ultrasound image denoising method, the algorithm shows it can extracte high matching virtual channel,and the denoising performance is enhanced.Meanwhile the image feature details are protected better, and the evaluation of the denoising performance are better than the classical method.In order to solve the problem of how to select the wavelet thresholding and the adjustment function in wavelet enhancement methord, and the shortage of lacking detail or edge. An image enhancement algorithm for medical ultrasound based on ICA and wavelet is proposed. Firstly, the algorithm extracts the feature of the medical image using ICA, and experiments show the feature have strong sense of hierarchy, sharp edges and rich detail. Then it will generate characteristic parameters which need to adjust daptively using Wavelet decomposition.At last it can enhance image combined with the theory of wavelet enhancement.Experimental results show that the algorithm improves image contrast and definition. Meanwhile the edge clarity and enhanced image has more detail.They correspond with the characteristics of human visual. The enhancement effect is better than the classical method.