Medical Image Retrieval Technology Based On Uncertain Location Graph

The development of medical images acquisition and storage technology has led to the rapid growth of the relevant data. According to the domain knowledge, we believe that finding similar images from the medical image database will significantly assist doctors in findng patients who may get the same disease. And it will also help doctors to make diagnoses by acquiring information from the previous diagnoses and result.However the similarity retrieval for medical image requires accuracy much higher than the normal images. There is no effieient feature model for medical image descritption and retrieval until now. And with the increasing number of medical image, current medical image retrieval methods all face to the time-consuming problem and there is no solution.In this paper, a novel model of uncertain location graph is presented for medical image modeling and similarity retrieval. According to the characteristics of medical image, we propose a novel method to model brain CT images to uncertain location graphs based on brain CT image textures. Then a scheme for Uncertain Location Graph similarity Retrieval (ULGR)is introduced. Furthermore, an effective index structure is applied to reduce the searching time.Experimental results reveal that this novel model functions well on medical images similarity retrieval with higher accuracy and efficiency.In the view of users are only interested in the most k similar results, we propose a medical image Top-k query method. With the result of ULGR, we proposed a model of association graph to represent the relationship of each pair of images. Further,a series of association measurements are proposed for reasoning which to reduce the number of image to image comparisons. Due to this measurement,a Top-k medical images query method is proposed and several walk strategies have been discussed. Moreover, the experimental results verify that the methods we proposed are efficient in practice.To sum up, the novel model of Uncertain Location Graph proposed in this paper functions well on medical images similarity retrieval with higher accuracy and efficiency. And the medical image Top-k query method effectively reduces the image-to-image comparison procedure.