Geri Dön

İnterpolasyonsuz bilgisayarlı tomografi

İnterpolation-fire computerized tomography

  1. Tez No: 46273
  2. Yazar: AHMET GÖNÜLLÜ
  3. Danışmanlar: Y.DOÇ.DR. SEDEF KENT
  4. Tez Türü: Yüksek Lisans
  5. Konular: Elektrik ve Elektronik Mühendisliği, Electrical and Electronics Engineering
  6. Anahtar Kelimeler: Bilgisayarlı tomografi, Computed tomography
  7. Yıl: 1995
  8. Dil: Türkçe
  9. Üniversite: İstanbul Teknik Üniversitesi
  10. Enstitü: Fen Bilimleri Enstitüsü
  11. Ana Bilim Dalı: Belirtilmemiş.
  12. Bilim Dalı: Belirtilmemiş.
  13. Sayfa Sayısı: Belirtilmemiş.

Özet

ÖZET N boyutlu fonksiyonların (N - 1) boyutlu izdüşümlerinin görüntüleme teknikleri ile birlikte kullanılması tıpta ve endüstride görüntüleme olayına yepyeni boyutlar kazandırmıştır. Bu uygulamalardan biride Bilgisayarlı Tomografidir. Görüntülenmesi istenen cismi çevreleyecek biçimde değişik açılardan ardarda yapılan taramalar sonucu elde edilen X - ışın demetine ait zayıflatma değerleri elde edilmektedir. Bilgisayar yardımı ile matematiksel bir çözümleme sonucu taranan cismin görüntüsü yeniden oluşturulmaktadır. Görüntü oluşturmanın temel esası projeksiyon datası olarak bilinen bir kaynaktan çıkan ışınlar ile bir cismin aydınlatılması sonucu elde edilen gölge diyagramından cismin bir kesitinin görüntüsünü oluşturmaktır. Bu tezde iki yöntem geliştirilmiştir. Her iki yöntemde de her hangi bir interpolasyon kullanılmamıştır. Projeksiyon - Dilim teoremine göre, işaret uzayındaki bir projeksiyonun tanımlanması Fourier uzayındaki bir dilimin tanımlanmasına karşı gelir. Birinci yöntem Projeksiyon - Dilim teoremi ile birlikte ters Radon dönüşümünü kullanılarak geliştirilmiştir. Ayrıca bu yöntemde görüntünün yeniden oluşturulmasında görüntünün kalitesi üzerinde etkili olan filtre fonksiyonları ele alınmış, sonuç görüntüler üzerinde ise iki filtrenin etkisi kısaca incelemiştir. Geleneksel X - ışınlı tomografi algoritmalarında, Projeksiyon - Dilim teoremi uyarınca Fourier uzayı, projeksiyonların Fourier dönüşümleri ile doldurulduktan sonra görüntünün elde edilmesi için iki boyutlu Fourier dönüşümü almak gereklidir. Kutupsal koordinat sistemindeki örnekler Fourier domeninde herhangi bir interpolasyon yöntemi kullanılarak kartezyen koordinat sistemine eş aralıklı noktalara yerleştirilebilir. Tezde kullanılan ikinci yöntemle Fourier domeninde herhangi bir interpolasyon yapmak yerine, oluşturulan projeksiyonlardan eş aralıklı kartezyen noktalara düşen örnekleri seçen bir algoritma geliştirilmiştir. Her iki yöntemle geliştirilen görüntü oluşturma algoritmaları çalışma hızı olarak olduça iyi sonuçlar vermiştir. Görüntü kalitesi açısından orijinal cisim ile karşılaştırılmış ve Ortalama Karesel Hataları (Mean Square Error - MSE) ve İşaret/Gürültü oram kriterleri için doyurucu sonuçlar elde edilmiştir V

Özet (Çeviri)

SUMMARY INTERPOLATION - FREE COMPUTERIZED TOMOGRAPHY An important problem in image processing is to reconstruct a cross section of an object from several images of its transaxial projections. A projection is a shadowgram obtained by illuminating an object by penetrating radiation. Fig 1 shows a typical method of obtaining projections. Each horizontal line shown in this figure is a one dimensional projection of a horizontal slice of object. Each pixel on the projected image represents the total absorption of the X-ray along its path from the source to detector. By rotating the source - detector assembly around the object, projection views for several different angles can be obtained. The goal of image reconstruction is to obtain an image of a cross section of the object from these projections. Imaging systems that generate such slice views are called CT ( Computerized Tomograhpy ) scanners. In obtaining the projections, we lose resolution along the path of the X - rays. CT restores this resolution by using information from multiple projections. Therefore, image reconstruction from projections can be viewed as a special case of image restoration. X-rays Fig 1. A typical method of obtaining projections VIFor X - ray CT scanners, a simple model of the detected image is obtained as follows. Let f (x,y) denote the absorption coefficient of the object at a point (x,y) in a slice at some fixed value of z axis (Fig. 1). Assuming the illumination to consist of an infinitely this parallel beam of X - rays, the intensity of the detected beam is given by I = I0 exp [ 1 f(x, y)du] (1) where is I the intensity of the incident beam, L is the path of the ray, and u is the distance along L ( Fig 2 ). Defining the observed signal as g = In g(I0/I) we obtain the linear transformation = g(s,6) = J f(x,y) du -oo<s<qo, 0 < 9 < tc (2) (3) Fig 2. Projection imaging geometry in CT scanning where (s,0) represent the coordinates of the X-ray relative to the object. The image reconstruction problem is to determine f(x,y) from g(s,9). In practice we can only estimate f(x,y) because only a finite number of views of g(s,0) are available. Fig 3. shows an X - ray computerized tomoghrapy VII( CT ) system. X-ray CT systems are used in medical imaging and nondestructive testing of mechanical objects. Fig 3. An X-ray CT scanning system The techniques that exist for reconstruction fall into two basic classes one in which the reconstruction is performed in signal space and one in which it is performed in Fourier space. While any techniques can, of course, be interpreted and analyzed in either space or in both together, most techniques are more easily implemented in one space than the other. Whether implemented in signal space or Fourier space, the reconstruction algorithms can be conveniently interpreted by means of a straightforward and interesting theorem which we refer to as the projection - slice theorem. In essence this theorem states that the Fourier transform of a projection is a slice of the Fourier transform of the projected object. A projection is a mapping of an N - dimensional function to an (N-l) dimensional function obtained by integrating the function in a particular direction. For example, p (x,) given by x2 +00 I PX2(x,)“lJ -00 f(x x7)dx. (4) is an example of a projection of two - dimensional function f(x x ) onto one dimension. VIIIFor the general case, we define a projection as follows: Let f(x) denote an N - dimensional function and let u denote a new set of coordinate where x = uA (5) and A is an orthogonal transformation. Then a projection onto the hyperplane ( u,, u7, u-_, >u-+],...u ) is defined as +00 P u/ul' u2' ui-l 'Ui+P -un) =. J f(-u A > d ui (6) The coordinate axis u-, which is normal to the hyperplane onto which f(x) is projected, will be referred to as the projection axis. For N = 2, the matrix A is given by cos 9 sin 9 A = -sin 9 cos 9 (7) Basically, the projection - slice theorem states that the (N-l) dimensional Fourier transform of a projection is a ”slice“ through the N- dimensional Fourier transform of f(x). Consider a projection for which the projection axis is one of the coordinate axes of f(x), for example, x,. Then p ( x...,0 is given by.+00 p x ( x2,...,xN) = J f(x) dxj (8) And its (N - 1) - dimensional Fourier transform is given by.+00 ”+°o I I dx^.d^ (9) PXj( 0)2,...,a>N) =J J px( x2,...fXN) exp[-j((D2x2+...+0)^)] Clearly, it can be seen that IXM“V,o>N)=F(a>lf...,a>N) |m =0 (10) 1 1 In other words, P (co,...,co ) is a slice of F(co,...,co ) defined by A - â JN 1 IN In utilizing projections for reconstruction, many of the algorithms involve computing the Fourier transform of the projections. The transform of each projections is a function of a set of continuous variables, but only a finite number of points from each Fourier transform can be computed and stored. Thus from the projections, only samples are available, in part because of the limited number of projections and in part because only samples of the Fourier transform on each slice can be obtained. The essence of the reconstruction problem, then, is to approximate all of Fourier space from its values on a discrete point set. Since specification of a projection in signal space corresponds to specification of a slice in Fourier space, if all the projections for a continuous range of angle 0 < 8 < % are known, the entire Fourier space is swept out and consequently the function is known exactly. As will be shown, image reconstruction from the projection can be thought of as the inverse Radon transform of the projection data pa(x'). Let us consider the 2 - D inverse Fourier transform operation of F (cox,00y). The estimated image function f(x,y) can be obtained by the inverse Fourier transform. f(x,y) = F2 [F(cox, coy ) ].+00.+00 = J J [F(cox,(Dv)]exp[j(Xtox + Ycov)] dcoxdoDv -00 -00 J J J = f(r,0 (11) If we write (üox, (öy) in Eq. (11) in polar coordinates (co,0), it can be written as r+oo J F(g>,0) exp [ jco (x cosG + y sinG ) ] | J | dco d6 (12) X/ 9 ? where (x cosB + y sinG) = x', co = v cox ~ + coy % coy = co sin0, cox=cocos0, 0 = tan~'[ coy / cox ] and | J | is the Jacobian The Jacobian is given as Jl = dcox / d co dcûy / 8 co 5cox / d 9 acoy / 50 cos 0 sin 0 -co sin 0 co cos 0 9 ? CO cos”0 + CO COS~0 I co I (13) By changing the limits of integration in Eq.(12) to 0 < 0 < n and -oo<co<oo, and replacing F(co,0) with Pq(co), Eq.(12) can be rewritten as r.+oo J 0 0 | co | Pq (co) exp [ jco (x cos0 + y sin0) ] dcod0 0 Where =J P0*(x') d0 (14) -+00 * I P0 (x') = J I co I P0 (co) exp [ jco x ' ] dco -1 = Fj [|co| Pe(co)] -1 F1 [|co|]*p0(x') (15) XIIn Eq.(15) the asterisk ( * ) denotes the 1 - D convolution operator. Note that pg (x') represents the filtered projection data. Chapter 1 is an introduction to X - ray Computerized Tomograhpy (CT) and a theoretical background for projection including projection - slice theorem. In this chapter, image reconstruction from projection is described and the previous studies are shortly reviewed. In chapter 2 a approximate reconstructions from a polar raster is defined and the Radon transformation and the inverse Radon transformation is described. At the end of the chapter the filter function is studied that is important in image reconstruction algorithms. In chapter 3 a new reconstruction algorithm is presented. In this algorithm we have tried to not use any interpolation such as angular, radial, bilinear, nearest - neighbor interpolation. Instead of to use an any interpolation on the Fourier space we could develop an algorithm that can select the samples are countered on the cartesian points with same intervals from the projections. At the further part of the chapter 3 another reconstruction algorithm is studied which is not used any interpolation. This algorithm performs Eq. (14). At the end of the chapter we can see the printouts of these algorithms. The software programs of these algorithms are given at the end of this thesis' s appendix. xn

Benzer Tezler

  1. Tomografik görüntü oluşturulmasında doğrudan fourier yöntemine yeni bir yaklaşım

    A Novel approach to direct fourier transform technique in tomographic image reconstruction

    A.FİLİZ BAYTAŞ

    Doktora

    Türkçe

    Türkçe

    1994

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİstanbul Teknik Üniversitesi

    Nükleer Mühendislik Ana Bilim Dalı

    PROF. DR. MELİH GEÇKİNLİ

  2. Object-oriented motion abstraction

    Nesneye yönelik hareket soyutlaması

    BİLGE ERKAN

    Yüksek Lisans

    İngilizce

    İngilizce

    1993

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİhsan Doğramacı Bilkent Üniversitesi

    Bilgisayar Mühendisliği Ana Bilim Dalı

    PROF. DR. BÜLENT ÖZGÜÇ

  3. Animating facial images with drawings

    Yüz görüntülerini çizimler yardımıyla canlandırma

    GAMZE DİLEK TUNALI

  4. Su darbeleri sonucunda oluşan kavitasyon olayının karakteristikler yöntemi kullanılarak incelenmesi

    Başlık çevirisi yok

    AYŞE KAYGISIZ

    Yüksek Lisans

    Türkçe

    Türkçe

    1989

    İnşaat MühendisliğiDokuz Eylül Üniversitesi

    Hidrolik Ana Bilim Dalı

    DOÇ.DR. M. ŞÜKRÜ GÜNEY