Font Size: a A A

Research On Clinical Image Analysis In Digital Diagnosis

Posted on:2009-03-12Degree:DoctorType:Dissertation
Country:ChinaCandidate:G B ChengFull Text:PDF
GTID:1118360272962138Subject:Biomedical engineering
Abstract/Summary:PDF Full Text Request
In recent years,more and more modem medical imaging equipments have been used in hospitals.Medical imaging technologies are playing more and more important roles in clinical diagnosis and medical research.Multi-dimension real-time screening of medical imaging is becoming necessary in hospitals.However,for better aiding clinical diagnosis,high volume medical information must be processed,integrated and mined effectively,which makes intelligent analysis of multi-dimension medical data necessary in daily clinical work.Thus how to combine clinical diagnosis with current research of medical radiology is very important,which can be shown in two reasons mentioned below.Firstly,high resolution images,ultra-fast scanning speed and a broad range of clinical applications are limited by human resource in hospitals.For example, multi-slice CT captures significantly much larger data sets and yields much sharper, more detailed images than traditional CT.But restricted by some well-known reasons, radiologists need to review large image sets quickly and easily with the help of sophisticated medical imaging process technology that can get valuable information from advanced multi-slice images,to avoid missing important diagnostic information in limited times.Secondly,new medical imaging technology should become popular not only in diagnostic department,but in clinical department.For example,now four-dimensional (4D) ultrasound lets women and doctors look at facial features and watch the growing baby move.With 3D ultrasound,a volume of echoes is taken,stored digitally,and shaded to produce life-like images of the fetus.A 4D ultrasound takes the images produced by 3D ultrasound and adds the element of movement.Now,the life-like pictures can move and the activity of the fetus can be studied.To doctors,4D reveals more detail about fetal health and small movement.Just as a pediatrician begins an exam by observing a newborn,doctors assess the fetus from head to toe on screen.By watching him or her shift position and breathe,doctors can check for problems.But restricted by practical situations in most hospitals,the above mentioned functions can not be utilized by clinical doctors.To a large extent,this has limited popularity and effect of new medical imaging analysis technologies in clinical use much.For this purpose,the thesis focuses mainly on research of practical application of medical imaging analysis in clinical environment.Main topics are listed below.First,a new medical image segmentation algorithm - marker-controlled watershed segmentation using the edge-detecting algorithm based on Generalized Fuzzy Operator is proposed in the article.Usually,traditional watershed segmentation is implemented based on the edge-extracted image from the original image.However, the conventional edge-detecting algorithms have not the prior information as well as shape restriction.Because of the changing pathology and the inherent property of fuzzy imaging data,the conventional edge-detecting can hardly guide watershed algorithm converging into the correct object contour;the resultant watershed algorithm is subjected to image noise and over-segmentation.In this paper,we propose to improve marker-controlled watershed segmentation using the edge-detecting algorithm based on Generalized Fuzzy Operator.Judginging from the results,the proposed method is very adaptive to the segmentation of the clinical images.Second,segmentation of medical images based on Gibbs morphological gradient and distance map(DM) Snake model is proposed to find right contours of objects when processing medical images with noises and pseudo-edges.To begin with,Gibbs morphological gradient is introduced.Then segmentation based on Gibbs morphological gradient and distance map snake model is proposed.Experimental results show that the new segmentation algorithm of medical images,based on Gibbs morphological gradient and distance map snake model can suppress noises and pseudo-edges.A conclusion can be drawn that advantages of the method are robust to image noise,easy to be implemented in clinical image segmentation with only a few user interventions.Third,a new method-region-based image display and enhancement for improving image display quality is introduced.It is useful to improve image display quality and enhance image based on different regions in the image,as similar human organs are often clustered.Different human organs are separated by image segment method.Then each area indicating different human organs are mapped to 1-256 gray level with different transformation functions.Thus,more than 256 levels of gray information can be show for clinical diagnosis.Region-based image enhancement is also very useful and is discussed in the paper.Fourth,acceleration research of the method mentioned above is done based on Graphics Processing Unit(GPU).Now GPU is used as general-purpose processing unit in many other fields,because it can become a general-purpose computing engine without compromising its performance.The combination of a data parallel engine with more of the general-purpose flexibility of a traditional CPU offers a powerful model for our method,which consists of a mix of irregular matrix math and other logic.Thus,region-based image enhancement for improving image display quality can be run in real-time in clinical environment.All in all,with the rapid development of modem medical imaging equipments, how to use intelligent analysis of multi-dimension medical data in daily clinical work is becoming more and more crucial.Some useful methods are studied especially based on clinic questions proposed and tested in the paper.
Keywords/Search Tags:Image Segmentation, Gibbs Random Field, Medical Image Processing, Image Enhancement, Graphics Processing Unit
PDF Full Text Request
Related items