Medikal üç boyutlu görüntüleme ve bir uygulama

Abstract

Bu teze konu olan üç boyutlu (3B) görüntüleme birçok mühendislik alanında uygulama bulan ve güncel olarak üzerinde araştırmalar yapılan aktif bir konudur. Bu tezde üzerinde çalışılan ; üç boyutlu görüntülemenin tıbba olan katkısı, doktorlara ve uzmanlara gerek teşhiste gerekse tedavide sağladığı kolaylık ve doğruluktur. Giderek yaygınlaşan ve hastanelerdeki Bilgisayarlı Tomografi (Computer Tomography - CT), Magnetik Rezonans Tomografisi (Magnetic Rezonans Imaging -MRI ) gibi medikal görüntüleme cihazlarında kullanılan üç boyutlu görüntüleme programları her zaman daha fazla imkanlar sunarak doktorların ve tıbbın hizmetinde bulunmaktadır. Çalışmalarımızda CT'den alınan 2 boyutlu kesitler tarafımızdan geliştirilmiş olan bir algoritma ile bir leştirilmiş ve hafızada üç boyutlu bir resim yaratılarak bunun ekranda gölgelendirme ile üç boyutlu gözükmesi sağlanmıştır. Algoritmalarımız Sun Sparc 1+ iş istas yonunda (Workstation) çalıştırılmış ve yeterli hıza ulaşılmıştır. Hızla ilgili kriterler tezde açıklanacaktır. Ayrıca segmentasyon, kesme, döndürme, histogram analizi, animasyon gibi görüntü işleme programlarıda gerçeklen- miştir. Programlarımızda farklı gölgelendirme, döndürme ve segmentasyon methodları denenmiş ; en iyileri tesbit edilmiştir. Bu çalışmada 1. bölümde 3B görüntülemeye genel bir giriş yapılarak problemler tanıtılmış ve tezde yapılanlardan kısaca bahsedilmiştir. 2. bölümde, üç boyutlu görüntülemenin tarihçesi anlatılmıştır. 3. bölümde ise üç boyutlu görüntülemenin yararları, kullanım alanları açıklanmıştır. 3 boyutlu görüntülemede kullanılan metodlar bölüm 4 'de açıklanmıştır. Bölüm 5 'de ise tezde kullanılan matematiksel yöntemler ve programların mantığı açıklanmıştır. Bölüm 6' de ise tezdeki programlardan elde edilen sonuçlar verilmiş ve sonuçlar açıklanmıştır. Programın geliştirilmesi için olan önerileri ve üç boyutlu görüntülemenin geleceği bölüm 7 'dedir. Ekler ise üç bölüm halindedir. Yazılan programlar ise Ek A'da sunulmuştur. Ek B'de Sun Sparc 1+ hakkında teknik bilgi verilmiştir. Ek C'de tezde geçen yabancı terimler açıklamaları ile verilmiştir.

Three Dimensional (3D) Imaging is an update subject in medicine because of it's diagnostic, therapueutic, teaching and research values. Research for 3D Imaging are being done to obtain better resolution and more speed for image processing while 3D Imaging programs are being used in hospitals and medical centers. Utility of the 3D imaging is incresing daily with the new developments in Computer Graphics and Fast Computers like workstations and several different images. Utilities can be presented to the specialists and clinicians for treatment and diagnosis by using digital image processing techniques as segmentationd, rotation around axis, clipping, artifical colors for different tissues, low pass filters, huge storing capabilities and etc. A well trained radiologist is able to mentally visualize an entire 3D image of the object from a series of cross sections of a target volume. In actual practice, it is often diffucult to accurately visualize the 3D surface of the object, especially when the configuration of the structure is complex. 3D processing of medical data is a diagnostic technique for clearfying the 3D relationships of complex lesions and anatomical structures. Computer Graphics allows this to be obtained as digitized information and has made it possible the full scale of pseudo 3D displays in the medical field. Clinical applications for 3D displays have been developed mainly in the following areas : Diagnostic support, simulations for surgical operations and therapy plannig. By using the 3D displays it is possible to diagnose morphological abnormalities which impossible to do accurately from individual slices of images, as well as to confirm 3D positional relationship between organs. The second area has an importance for surgical operations and therapy planning. During brain surgery this technique allows surgeons before surgical operations to understand the true location of the lesions. These utilities make the surgical operations safer than they were and this point of view, it is very easy to understand that why we need high speed processing of the 3D images. Orthopedists and plastic surgeons also use CT images to construct 3D images of bones and soft tissues, especially in the face. XI In the therapy planning it is very important to define the true location of lesion which will be applied radiation. Another important point is, to calculate the radiation dose distrubition within the irradiated volume. 3D displays allow this activities during the radiation treatment. In the area of medical education, 3D images simplfy the diffucult anatomical explanations in the textbooks by converting explanatory diagrams into computer images. To obtain 3D images there are two main methods : volume rendering and surface rendering. In volume rendering methods we have to calculate volumetric data and use it to show the images. In the other method we calculate only the surfaces of the objects from the individual slices. In volume rendering methods, more data have to used to obtain the result and the process time is more than surface rendering techniques. Regardless of these problems, volume rendering methods allow us to see the all volume in clipping and segmentation process because it contains all volume data which belongs to 3D object. Even surface rendering methods are fast and need less data, they don't contain volume information. In our programs we had to obtain segmented and clipped objects so we had to use volume rendering methods. In our programs, we only used surface information to provide fast rotating because of the short processing time of this kind of images. During our studies we realized the difference between this two methods. Even we can't clipped or segmentated the object which is obtained by using surface rendering methods, the rotating rate is at least 10 times more than the object's rotating rate which is obtained by using volumetric data. At the beginning of our studies we used some animation datas that we created them in the computer by using programs. After we write our programs and tested them by using this animation datas, we started to use real medical data. Our first medical datas aren't enough to obtain a real medical data (25 slices), so we increased the number of the slices by using an interpolation algorithm (168 slice). Images from the CT and MRI computers have got gray levels and we have to use this gray level to obtain our objects. In CT images gray levels represent the attenuation coffiecent of the x-rays which are passing the XII body and in an original image there are 4096 gray level so because of the screen limits and anatomical structure of the human eyes (human eye can't differ the difference between more than 20 gray level) we have to create a window in the gray level spread. Our window contains 256 gray level to obtain more detail and accurate results. This window can be adapted any part of 4096 gray level spread and allow us to see the information in the window. Our screen is two dimensional so if we want to create 3D images in the screen we have to use some computer graphic algorithms like shading and hidden surfaces. We used an algorithm for shading which is called 26 neighbourhood gray level gradient shading algorithm. We used some different shading algorithms to observe the difference between the results and we decided that 26 neighbourhood gray level gradient shading was the most appropriate. This algorithm is a version of the Phong Shading Algorithm. Two dimensional slices are located in the memory to create a 3D array. Our 3D object is located in a constant x,y,z coordinate system, when our observer coordinate system x',y',z' is rotating around the object. We prefered this rotating system because we think that rotating time would be less. The direction of the light, which is used for shading, is given on the x,y,z coordinate system. Direction of the light can be changed or adjusted automatically. Segmentation, is a very important aspect of the 3D imaging study. If the segmentation program is good enough and the images contain the neccesary data ; you can observe different tissues from images. For example it is very diffucult to observe vessel information from a CT image, but it is quite easy from a MRI or an Angiography image. By using some complex algorithms, it is possible to find the cortex and ventricular structure of the brain. Gray level windowing is a kind of segmentation but it is not enough to catch desired tissues. With segmentation, all skin information can be eliminated in order to see the bone structure of the head. Another application would be to eliminate the bone and see the brain only. At this point of study, we can clip the image and see the inside of the brain. This kind of utilities are very usefull for brain surgeons because they can find the true location of the foreign body and they provide accurate information for the operation. Xlll body and in an original image there are 4096 gray level so because of the screen limits and anatomical structure of the human eyes (human eye can't differ the difference between more than 20 gray level) we have to create a window in the gray level spread. Our window contains 256 gray level to obtain more detail and accurate results. This window can be adapted any part of 4096 gray level spread and allow us to see the information in the window. Our screen is two dimensional so if we want to create 3D images in the screen we have to use some computer graphic algorithms like shading and hidden surfaces. We used an algorithm for shading which is called 26 neighbourhood gray level gradient shading algorithm. We used some different shading algorithms to observe the difference between the results and we decided that 26 neighbourhood gray level gradient shading was the most appropriate. This algorithm is a version of the Phong Shading Algorithm. Two dimensional slices are located in the memory to create a 3D array. Our 3D object is located in a constant x,y,z coordinate system, when our observer coordinate system x',y',z' is rotating around the object. We prefered this rotating system because we think that rotating time would be less. The direction of the light, which is used for shading, is given on the x,y,z coordinate system. Direction of the light can be changed or adjusted automatically. Segmentation, is a very important aspect of the 3D imaging study. If the segmentation program is good enough and the images contain the neccesary data ; you can observe different tissues from images. For example it is very diffucult to observe vessel information from a CT image, but it is quite easy from a MRI or an Angiography image. By using some complex algorithms, it is possible to find the cortex and ventricular structure of the brain. Gray level windowing is a kind of segmentation but it is not enough to catch desired tissues. With segmentation, all skin information can be eliminated in order to see the bone structure of the head. Another application would be to eliminate the bone and see the brain only. At this point of study, we can clip the image and see the inside of the brain. This kind of utilities are very usefull for brain surgeons because they can find the true location of the foreign body and they provide accurate information for the operation. Xlll At the beginning chapter of this study an overview of the 3D imaging is given. In the second chapter, the history of the 3D imaging is written. In the third chapter, the utilities of the 3D imaging and the interested areas are being explained. Three dimensional methods are explained in the chapter four. Our methods and formulas with our three dimensional rendering techniques are in the fifth chapter. In the sixth chapter, all results of our study are given and explanied. There are future adviceses for our 3D programs and an overview for the future of the 3D imaging in chapter seven. In the additional part A, you can find more technical information about Sun Sparc 1+ and some utilities for compailing C programs, writing in a vi editor. In additional part B, some of the terms in the thesis are explained. In part C, our programs are given with their explanations.