Research On CT Image Quality Evaluation And Testing Performance Phantom

Computer tomography is the most important support method of medical diagnosis nowadays. From the view of technical or physical point,quality of the image is the CT decisive parameter,and the evaluation of CT image quality must have an objective basis. Evaluation image quality of the CT device by parameters:spatial resolution, density resolution, noise, layer thickness, CT value linear, uniformity and so on, and these need to use test phantom for measuring.In the evaluation of quality of CT image, the spatial resolution is often first considered. The detection methods of spatial resolution include:line pair method, circular method, modulation transfer function method (Modulation Transfer Function, referred to as MTF) etc.. Method of line pair and hole method belong to the subjective measurement method, measured directly by the use of periodic structure (line pair, circular), consistent with the defining of spatial resolution in physics with an intuitive, convenient characteristic; but easily influenced by subjective factors, so MTF method is often adopted in the technology test of CT. MTF often uses two kinds of test phantom, one is by scanning a homogeneous medium metal ball or fine metal wire to get point spread function,the other one is by scanning high contrast interface to get edge spread function, and then use FFT of the line spread function. In point spread function method,the manually selected method is mainly used for getting center position.To manually select the center point position differ from man to man, leads to the final result can not be unified, in using a process actually, there will be some confusion, such as different testers to get different results, unable to judge which result is more accurate. In order to get the uniform result, an algorithm is needed to avoid the none stable manual selection result and not affected by the influence of noise.According to the characteristics of point spread function, corresponding to the area should be a circular area,in this paper, by means of image processing method, extract the circular area, then use related mathematical transformation to find the center point position. In the edge spread function method, due to the noise, non uniformity, variability of data,algorithm limitation and so on, errors of measurement will be inevitably produced. Mean of experiments is a good method to decrease the error, however it will raise a requirement on phantom images: the goal line between two kinds of materials must be vertical or horizontal. In this article, we do transform on phantom images, to achieve conditions for direct average, simplify the clinical scanning, do convenience for the assessment of CT system.Image noise directly influences low contrast resolution of CT system, bring inconvenience and error for further study or use, therefore, CT image should be preprocessed in order to reduce the noise. In image preprocessing, if assume noise types are known, blind to do denoising directly, CT image often can not reach the good effect, because of the treatment effects in different denoising methods for different types of noise vary widely.Different noise types should be used for the corresponding noise algorithm, so the recognition of noise types before image treatment is very necessary, and is helpful to the study of image denoising algorithm,too. The noise type identify methods, such as analysis wavelet coefficient histogram energy, jump probability and window width at golden section point of curve fitting and so on, these methods are more accurate for noise type identification. On this basis, put forward a kind of more simple noise type identification method of CT image. The method based on wavelet decomposition, by analysis the signal to noise ratio of frequency coefficient histogram and integral of the corresponding curve fitting, identify CT image noise types efficiently (for example, based on the Gauss noise and salt&pepper noise) so as to provide guidance for the design of denoise method, improve preprocessing effect of CT image.In order to ensure the effective application of CT equipment in hospital, equipment must be regularly tested, and phantom is a necessary tool. Along with more and more CT equipment users, the demand of test phantom will grow. The phantom Introduced from abroad are expensive, limit its popularization in primary hospital greatly. In this paper, based on the production of the common phantom,we made practical and cheap CT detection phantom. It can detect the conventional image quality parameters of CT equipment, including spatial resolution, density resolution, noise, layer thickness, CT value linear, uniformity and so on.