Medical Ultrasound Image Enhancement Algorithm

In recent years, due to its unique real-time performance, non-destructive property, low cost, high sensitivity and other advantages, medical ultrasonic image is widely used in clinical diagnosis. When it is applied in clinic, it is usual to extract some information of the specific organs and areas, and an indispensable mean for this is image enhancement. We can extract the parts which we are interested in through image enhancement from an ultrasonic image,or find out the location of lesions and its shape. It decides the accuracy of the clinical pathologic analysis, diagnosis and treatment. However, because of the random disturbance of electronic devises in ultrasonic imaging system, and the effect of the environment, in the process of acquiring image, there will be some noises and distortions which may influence image quality. So, image enhancement must be taken to improve image quality. In course of ultrasonic image enhancement processing, by the means of the traditional images enhancement algorithms, compares the advantages and disadvantages of each method, and the three new algorithms were figured out.1. The algorithm based on soft morphology was proposed according to the weaknesses of traditional morphology algorithm. The bright features of ultrasonic images were extracted by multi-scale soft morphological filters. Based on the scale attributes of these features, and the demands of local contrast enhancement, the points whose were gray-values greater than threshold were processed, and the bright features were added to these points. The experimental results showed that the method was more effective, less sensitive to noise and preserving processed images more accurately.2. A weighted contrast enhancement method of medical ultrasound images based on similarity estimation was proposed. This method firstly calculated the similarity of the processing window around a pixel and matching template. According to the relationship between the similarity and each parts of ultrasound image, the matched images were enhanced by local enhancement approach when it satisfied with correlation condition. The weighted average method for multiple images was used to obtain the final enhancement results. Experiment results with this method showed better contrast resolution, better edge feature detection than the original, and also preserved the structure without losing useful clinical information. In this respect, this method remedied the shortage of traditional enhancement algorithms, but the matching template was still selected manually.3. An adaptive edge enhancement method based on histogram matching for ultrasound images was presented. Because of the speckle noises of ultrasound images, traditional edge enhancement methods did not work well for ultrasound images. The proposed method was to seek the regions whose edges needed to be enhanced and ruling out the error edges caused either by speckle noises or the reverberation artifacts. A region covered an edge in original image has been selected as the match template first. Then the similar values between pending image block and the match template were calculated according to local histogram matching. And the local image block was enhanced when its similar value reached on a certain level. Furthermore, the ultimately enhanced result was achieved by the method of multiple images weighted average. The experiments results showed the proposed method can leave speckle unchanged and enhance tissue boundaries.