## ARTICLE NO. IV960489 Shape from Radiological Density (1994)

### BibTeX

@MISC{Poli94articleno.,

author = {Riccardo Poli and Guido Valli},

title = {ARTICLE NO. IV960489 Shape from Radiological Density},

year = {1994}

}

### OpenURL

### Abstract

detected in each slice by means of surface fitting or interpo-In this paper we propose a strategy to solve the problem of lation methods [7–15]. Surface fitting approaches can also recovering the 3-D shape of anatomical structures from single be used for the recovery of the approximate shape of an X-ray images, i.e., the problem of Shape from Radiological anatomical structure from a small set of tomographic or Density (SFRD). In order to overcome the noninvertibility of radiographic images when the boundaries of such a structhe process of image generation, we formulate a minimal set ture or other surface landmarks (obtained, for example, of physical assumptions that are used to constrain SFRD and by matching fiducial points in different views) are availto transform it into a well-posed problem. Our shape recovery able [16–19]. strategy requires the solution of four problems: (a) linearization of the process of X-ray image generation, (b) image segmenta- The above-mentioned methods use the output data of a tion, (c) estimation of a map of the local thickness of each segmentation or a feature-matching algorithm as geometric anatomical structure of interest, and (d) recovery of the 3-D constraints for the recovery of 3-D shape. However, in the shape of each structure from its boundaries and thickness map. case of X-ray projective imaging, an additional, important In this paper we assume that problems (a) and (b) have already source of 3-D information is available: the selective absorpbeen faced, and propose a solution for problems (c) and (d). tion of X-ray photons by the different tissues being imaged. Experimental results on synthetic images, X-ray images of Such a source of information is used, for example, in phantoms, and real radiograms are reported. © 1997 Academic Press computed tomography for the reconstruction of a density image from a complete set of projections on the grounds 1.

### Citations

459 | On active contour models and balloons - Cohen - 1991 |

237 | Regularization of inverse visual problems involving discontinuities - Terzopoulos |

231 |
Finite Element Procedures in Engineering Analysis
- Bathe
- 1982
(Show Context)
Citation Context ... quadratic function E˜ ( f ) a scaling factor only. In our experiments (if not otherwise that is discretized with the finite element method (FEM) stated) we have resolved some of these ambiguities by =-=[45]-=-. Basically, FEM consists in dividing the domain D into arbitrarily setting the z position of the center of mass of a set of small rectangular subdomains and hypothesizing the model and estimating the... |

144 |
The computation of visible-surface representations
- Terzopoulos
- 1988
(Show Context)
Citation Context ...h terms mathematically, a reference system and a mathematical representation for the surface S must first be chosen. In this work we have adopted a cylindroidal model of S, similar to the one used in =-=[38, 40]-=-, which can be represented by the parametric equation x(u, v) � ve 1 � f (u, v) cos(u)e 2 � f (u, v) sin(u)e 3, where f (u, v) is a twice differentiable function, periodic with respect to u, (u,v)�D�[... |

30 |
Interactive display and analysis of 3-D medical images
- Robb, Barillot
(Show Context)
Citation Context ...prings and ui is such that els such as the one described in [15]. 2.2.1. Evaluation of Eboundary. The boundary of a f (ui, vi) cos ui ��maxu f (u, vi) cos u if ri � 0, minu f (u, vi) cos u otherwise. =-=(2)-=- given structure in the image plane is the projection of a 3-D curve belonging to the surface of the structure, termed apparent contour. The apparent contour does not necessar- 2.2.2. Evaluation of Ed... |

30 |
The Physics of Radiology
- Johns, Cunningham
- 1983
(Show Context)
Citation Context ...herefore, in theory, the local thickness of each structure of interest can be recovered up to a scaling factor only. However, in biological tissues, �h usually has a value in the range 0.21–1.05 cm�1 =-=[36]-=-. This provides a lower and upper limit FIG. 8. Two views (a) and (b) of the shape of the two vials shown in Fig. 6a.SHAPE FROM RADIOLOGICAL DENSITY 367 recovery is aimed at integrating all the infor... |

25 | Algebraic reconstruction in CT from limited views - Andersen - 1989 |

17 |
High-speed three-dimensional X-ray computed tomography: The dynamic spatial reconstructor
- Robb
- 1983
(Show Context)
Citation Context ...hown in Fig. 6a. y0) � 0, and as ending point a generic point (x, y) inside �h, the previous equation transforms into s h(x, y) �� l 0 �s h(x(t), y(t)) � n(t) dt � 1 �h� l �I(x(t), y(t)) � n(t) dt, 0 =-=(1)-=- where [x(t), y(t)] is an arbitrary curve such that [x(0), y(0)] � (x0, y0) and [x(l), y(l)] � (x, y), and n(t) � [x˙(t), y˙(t)] is the tangent vector of such a curve. This equation provides a method ... |

16 |
Knowledge-driven ultrasonic three-dimensional organ modeling
- Brinkley
- 1985
(Show Context)
Citation Context ...d/or v, and n � x u � x v/�x u � x v� is the surface normal. Substitution of these equations into Eq. (6) leads to an express for E surface( f ) of the form E surface( f ) ��� D I surface( f ) du dv, =-=(7)-=- where I surface( f ) is a complicated, nonquadratic, differential operator as in [15]. 3. IMPLEMENTATION Some of the details of our implementation of the methods for thickness estimation and shape re... |

13 |
Fast direct display of volume data for medical diagnosis
- Ylä-Jääski, Klein, et al.
- 1991
(Show Context)
Citation Context ... cos u � r(x, y)]. u�[�,2�) This formulation for the densitometric constraints would not overconstrain the surface to be recovered. However, when a single radiogram is used, densitometric constraints =-=(4)-=- Edensity( f ) � 1 �d��f (uu(x, y), v(x, y)) � 2� xy sh(x, y)�(x, y) 2 � sin uu(x, y) ��f (ud(x, y), v(x, y)) � s h(x, y)(1 � �(x, y)) sin ud(x, y) where �(x, y) is the ratio between the thickness of ... |

9 |
Ritman, Extraction of leftventricular chamber from 3-D CT images of the heart
- Higgins, Chung, et al.
- 1990
(Show Context)
Citation Context ...and ud(x, y) are such that part and the other to the lower part of S, which force the surface to be more consistent with thickness data near the f (uu(x, y), v(x, y)) cos uu(x, y) � r(x, y) boundary. =-=(3)-=- According to this formulation, we can express the elastic � min [ f (u, v(x, y)) cos u � r(x, y)] energy of the springs which model the action of thickness u�[0,�) data as and f (u d(x, y), v(x, y)) ... |

5 |
Fomin Elementi teorii funktsij i funktsionalnogo analiza Moskow
- Kolmogorov, V
- 1976
(Show Context)
Citation Context ... already mentioned in Section ce( f ) in Eq. (7) can be approximated with its truncated 3.1, if the linear absorption coefficient is not known a priori, Taylor expansion I˜ surface(v, f, f0) about f0 =-=[44]-=-. The resulting the local thickness of a structure can be recovered up to approximation for Esurface( f ) is a quadratic function E˜ ( f ) a scaling factor only. In our experiments (if not otherwise t... |

2 |
Reconstruction of the left ventricle from two orthogonal
- Bai, Krishnaiah, et al.
- 1991
(Show Context)
Citation Context ...s are used in [30] for the reconstruction of ventricular shape from biplane angiocardiograms. Regularity of ventricular cross sections with respect to the two projection directions is hypothesized in =-=[31]-=-. In most of the above-mentioned methods the space that does not belong to the structure of interest is assumed to be empty of hypothesizing that such a structure is much denser than the other structu... |

1 |
A neural network archi- graphic screening
- Coppini, Poli, et al.
- 1993
(Show Context)
Citation Context ...x, y)) sin ud(x, y) where �(x, y) is the ratio between the thickness of the upper part of the surface and the total thickness: 2 � �,370 POLI AND VALLI sin uu(x, y) f (uu(x, y), v(x, y)) �(x, y) � . =-=(5)-=- sin uu(x, y) f (uu(x, y), v(x, y)) � sin ud(x, y) f (ud(x, y), v(x, y)) 2.2.3. Evaluation of E surface. If we imagine the surface S to result from bending a thin plate [42], so that a sort of deforme... |

1 |
Estimating the 3-D skele
- Kitamura, Tobis, et al.
(Show Context)
Citation Context ...l density (SFRD). The method presents the followingfeatures: (a) only a single X-ray image is required (even if the method is adequate for FIG. 1. Cylindroidal surface model. of X-ray images [24]. In =-=[25]-=-, the assumptions of elliptical cross section and constant density of vessels allow for the estimation of the shape of coronary arteries from two X-ray projections (biplane angiograms). The hypothesis... |

1 |
Three-dimensional segmentation and tons and transverse areas of coronary arteries from biplane angiointerpolation of magnetic resonance brain image
- Joliot, Mazoyer
- 1988
(Show Context)
Citation Context ...gine the surface S to result from bending a thin plate [42], so that a sort of deformed cylinder is obtained, its potential energy is proportional to the functional E surface( f ) �� S [2H 2 � K] dS, =-=(6)-=- where H and K are the mean and Gaussian curvatures of S, respectively. 2 From differential geometry [43] it is known that EN � GL � 2FM H � 2 �S 2 2 LN � M , K � �S 2 , dS ��Sdudv, where E, F, G and ... |

1 | Reconstructing the cross - Tran, Bahn, et al. |

1 | An network approach to implicit image information extraction from two analytical descriptor of three-dimensional geometry: Application to projections - Azhari, Sideman, et al. - 1987 |

1 | A simple approach to the 28 - Brevdo, Sideman, et al. - 1987 |

1 | Surface reconstruction of 3D objects in Trans. Med. Imaging 13 - Xu, Lu - 1994 |

1 | Improvement on dynamic 18–28. elastic interpolation technique for reconstructing 3-D objects from serial cross sections - Lin, Liang, et al. - 1990 |

1 | ReconstrucRoentgen-Video-Techniques for the Dynamic Studies of Structure and tion of 3D medical images: A nonlinear interpolation technique for Function of the Heart and Circulation - Chang, Chen, et al. - 1991 |

1 | Using deformable surface projections, Comput. Vision Graphics Image Process - Cohen, Cohen, et al. - 1989 |

1 |
Recovery of 3-D closed surfaces Thieme
- Poli, Coppini, et al.
- 1971
(Show Context)
Citation Context ...ver, with minor changes, the shape-recovery procedure described below could be used for other surface modwhere �b is the stiffness of such springs and ui is such that els such as the one described in =-=[15]-=-. 2.2.1. Evaluation of Eboundary. The boundary of a f (ui, vi) cos ui ��maxu f (u, vi) cos u if ri � 0, minu f (u, vi) cos u otherwise. (2) given structure in the image plane is the projection of a 3-... |

1 | Georg Thieme Verlag - Dumay, Gerbrands, et al. - 1978 |

1 | Medical Imaging Systems, Prentice Hall, Englewood dimensional reconstruction of myocardial contrast perfusion from Cliffs - Macovski - 1983 |

1 |
A line-integration based method gence
- Wu, LI
- 1988
(Show Context)
Citation Context ...t (x, y) is near � h, 0 otherwise. In other words, near the boundary of an anatomical structure the image gradient is proportional to the gradient of the thickness of such a structure only. Following =-=[35]-=-, if we hypothesize on the basis of A2 that the thickness maps sh(x, y) are differentiable to the second order with continuous second derivatives, thanks to Schwartz’s and Green’s theorems we have I(x... |

1 | A computa - Calamai, Coppini, et al. |

1 | Epicardial surface estimation from discontinuities - Young, Hunter - 1986 |

1 | Recovery of the 3-D shape of - Coppini, Poli, et al. - 1988 |

1 | Closed-form solution for physically based - Pentland, Sclaroff |

1 |
Digital Picture Processing, Vols. 1 and shape modeling and recognition
- Rosenfeld, Kak
- 1982
(Show Context)
Citation Context ... real radiograms are reported. © 1997 Academic Press computed tomography for the reconstruction of a density image from a complete set of projections on the grounds 1. INTRODUCTION of Radon’s theorem =-=[20]-=-. Unfortunately, when the number of projections available is small, image reconstruction becomes an extremely ill-posed problem that can be (approxThe recovery of the three-dimensional shape of anatom... |

1 | Algorithms for limited - Rangayyan, Dhawan, et al. |

1 |
Symmetry-seeking models view computed tomography: An annotated bibliography and a chaland 3D object reconstruction
- Terzopoulos, Witkin, et al.
- 1987
(Show Context)
Citation Context ...h terms mathematically, a reference system and a mathematical representation for the surface S must first be chosen. In this work we have adopted a cylindroidal model of S, similar to the one used in =-=[38, 40]-=-, which can be represented by the parametric equation x(u, v) � ve 1 � f (u, v) cos(u)e 2 � f (u, v) sin(u)e 3, where f (u, v) is a twice differentiable function, periodic with respect to u, (u,v)�D�[... |

1 | 3D reconstruction using a limited number 3rd European Conference on Computer Vision - Klifa, Lavayssière - 1994 |

1 |
tomosynthesis: A promising technique for angiocardio- ence
- Stiel, Stiel, et al.
(Show Context)
Citation Context ...diological density (SFRD). The method presents the followingfeatures: (a) only a single X-ray image is required (even if the method is adequate for FIG. 1. Cylindroidal surface model. of X-ray images =-=[24]-=-. In [25], the assumptions of elliptical cross section and constant density of vessels allow for the estimation of the shape of coronary arteries from two X-ray projections (biplane angiograms). The h... |

1 |
Computational vision and regular
- Lipschutz, Geometry, et al.
(Show Context)
Citation Context ...ined, its potential energy is proportional to the functional E surface( f ) �� S [2H 2 � K] dS, (6) where H and K are the mean and Gaussian curvatures of S, respectively. 2 From differential geometry =-=[43]-=- it is known that EN � GL � 2FM H � 2 �S 2 2 LN � M , K � �S 2 , dS ��Sdudv, where E, F, G and L, M, N are the first and second fundamental coefficients of S, respectively, and �S � �x u � x v� � (EG ... |