Network Of Medical Image Processing Key Technologies

With the breakthrough in computer technology, storage media andnetwork technology and the establishment of DICOM standard,PACS has been applied in clinical area step by step in recent years.This paper does researches both on medical image display andstorage. In medical display filed, pseudo-color coding, interpolationbetween layers and image segmentation are studied; in medicalstorage field, the CBIR, contented-based image retrieval, isdiscussed. At last, the way of desigh and implement of a medicalimage is given.The innovative points of this paper are summaried as follows:Firstly, an improved uncontinuous pseudo-color coding based onIHS apperceive color space is given, which is more appropriate forpeople's apperception compared with those based on RGB colorspace. In H plane, image background and object gray range areseparated. The background gray range is made as smaller as possibleand the object gray range as larger as possible. This method weakensthe background and stands out the frontground, which is propitiousfor doctor to get more useful information.Secondly, this paper proposes an improved contour interpolationalgorithm based on euclidean distance transform. The concept ofcontour box inscribed ellipse is proposed. The experiment resultsshow that the interpolation contours between extremely concavecontours are smooth and the transition is nature.Thirdly, an improved thorax contour interpolation algorithm isgiven, which extracts the contours of body and lungs. Comparedwith direct linear gray interpolation, it stands out the contoursinstead of blurring the object's boundary.Fourthly, a contented-based image retrieval method based on both feature integration and relevance feedback is proposed in this paper.Colour, texture and shape are three kinds of common features ofimages. Since each feature has advantage and disadvantage, thefeature intergration is adopted. Meanwhile, the relevance feedbackmechanism is used to enhance the effectiveness of retrieval byadjusting relevance weights. Experiment result shows that themethod based on both feature integration and relevance feedback hasbetter precisions and recalls than that of single feature extraction.Finally, Color feature of an image is one of the most common usedfeatures in content-based image retrieval (CBIR) systems.Combining the traditional color-histogram with some spatialinformation will greatly increase the accuracy of color features. Oneadvantage of color feature is that it is independent to the distortionof an image. However, with the introduction of the spatialinformation, this advantage will disappear. We proposed a newsymmetrical color-spatial method based on square loop. Experimentresults show that the method has good performance when 90 or 180degree rotation and flip. Its good efficiency makes it suitable forlarge numbers of images retrieval.