Local Transparency Control of Medical Slice Images and its Application to Slice-Slice and Slice-Volume Fusion

Yuta Fujimoto1, Saori Ojima1, Kyoko Hasegawa2, Tomoko Tateyama2, Susumu Nakata2, Yen-Wei Chen2, Satoshi Tanaka2

1 Graduate School of Science and Engineering, Ritsumeikan University, Japan 2College of Information Science and Engineering, Ritsumeikan University, Japan

Abstract In medical visualization, 2D slice images of human volume data are important as well as 3D volume rendering. The slice images are advantageous in focusing on regions of interest, while the volume rendering enables us to easily observe the whole region. In this paper, we propose a method to (1) create a transparent slice image, where its opacity is controllable by a user-defined transfer function and to (2) fuse a transparent slice image with another slice image or a volume three-dimensionally with correct depth feel. The transparent slice-slice fusion enables a doctor to observe multiple slices simultaneously. The transparent slice-volume fusion enables a doctor to focus on regions of interest on the slice, while observing the whole volume data. Opacities of the constituent slice images and a volume are both flexibly controllable with user-defined transfer functions. Our method is based on the particle-based rendering, which we recently proposed. The rendering is executable with interactive frame rates.

Keywords – medical volume data, slice image, volume-slice fusion, transparency control

1 Introduction In medical visualization, 2D slice images of human volume data are important as well as 3D volume rendering. The slice images are advantageous in focusing on regions of interest, while the volume rendering enables us to easily observe the whole region. In this paper, we propose a method to (1) create a transparent slice image, where its opacity is controllable by a user-defined transfer function and to (2) fuse a transparent slice image with another slice image or a volume three-dimensionally with correct depth feel. The transparent slice-slice fusion enables a doctor to observe multiple slices simultaneously. The transparent slice-volume fusion enables a doctor to focus on regions of interest on the slice, while observing the whole volume data. Opacities of the constituent slice images and a volume are both flexibly controllable with user-defined transfer functions. Our method is based on the particle-based volume rendering [1] and the particle-based surface rendering [2], both of which we recently proposed. In this paper, we use the term particle-based rendering by which we mean the both of the rendering algorithm.

2 Particle-Based Rendering The particle-based rendering [1,2] is a kind of transparent point-based rendering. This rendering enables us to easily execute fused visualization only by merging particles that describe constituent objects to be fused. Each fused object can be either a volume or a surface. It means that volume-volume fusion, volume-surface fusion, and surface-surface fusion are all possible through the same rendering procedure. In the particle-based rendering, the volume opacity becomes proportional to particle density. In other words, the volume opacity is controllable by properly tuning particle density distribution according to a given transfer function. On the other hand, the surface opacity, α, is controlled through the number of particles, n, on the whole surface as follows:

! =1! 1! sS

"

#$

%

&'n

, (1)

where, S and s are a surface area and a particle cross section, respectively. It should be noted that this formula holds under the assumption that αis uniform everywhere on the surface.

3 Local Opacity Control of a Slice Image In the current work, we extend applicability of the particle-based rendering to 2D slice images with non-uniform opacity. A slice is a kind of colored surface. Therefore, the conventional particle-based rendering [2] is applicable to a slice image, if its opacity is uniform. In this paper, we propose an extended method to create a transparent slice image with non-uniform opacity. We make the opacity locally controllable with a given transfer function by referring local pixel values. This makes local opacities of a 2D slice image controllable with the same procedure us the conventional 3D volume rendering. To realize the above-mentioned local opacity control, we divide a slice image into small square cells, in each of which a pixel value can be regarded as uniform (see Fig. 1.). Then, for each cell, we control its opacity according to the formula (1), assigning a proper local number of particles, n.

Fig.1 Local opacity control of a 2D slice image. A slice image is divided into small square cells. For each cell, its opacity is controlled with the local number of particles according to formula (1).

4 Results Fig.2 (left) shows an original CT image of a human body, while Fig.2 (right) shows a transparent 2D slice image with non-uniform opacity created by our method. In creating Fig.2 (right), the original 512 × 512 slice image is divided into 256 × 256

square cells. Each cell is uniformly sampled to generate a proper local number of particles that corresponds to desired opacity according to formula (1). In Fig. 2(right), opacity of the vertebra area is made larger so that the area can be highlighted, while other areas are made less opaque (more transparent). A transfer function used for the local opacity control is shown in Fig. 3.

Fig.2 An original slice image (left) and the result of the local opacity control (right).

Fig.3 The opacity transfer function used to create Fig.2 (right). Fig.4 shows transparent fusion of a horizontal and a vertical slice images. First, each slice image is sampled in the same way as Fig.2 (right). Next, the both particle sets are simply merged. Finally, the created larger particle set is used as rendering primitives of the particle-based rendering. Fig.4 demonstrates that the transparent feature of the slice images helps us to see the whole part of each image, not occluded by the other.

Fig.4 Fused visualization of a horizontal and a vertical slice images. Fig. 5 shows fusion of a transparent 2D slice image with the whole 3D volume. The way of executing the fusion is the same as Fig.4 except that particles for the 3D volume are generated according to the prescription of reference [1], making the particle density distribution proportional to the distribution of the voxel values. By using transparent slice-volume fusion, doctors can focus on regions of interest on the slice image, while observing the whole volume simultaneously.

Fig.5 Fused visualization of a 2D slice image with 3D volume rendering. In creating an image in Fig.5, 4.1 ×107 particles are used. Rendering is executed with an interactive FPS, using a standard laptop PC.

5 Conclusion In this paper, we proposed a method to (1) create a transparent slice image, where its opacity is controllable by a user-defined transfer function and to (2) fuse a transparent slice image with another slice image or a volume three-dimensionally with correct depth feel. Our method is based on the particle-based rendering that is a kind of transparent point-based rendering applicable to both volumes and surfaces. The slice-slice fusion helps a doctor to see the whole part of each constituent slice image, not occluded by the other. The slice-volume fusion helps a doctor to focus on regions of interest on the slice image, while observing the whole volume data. Opacities of the constituent slice images and volume are both flexibly controllable with user-defined transfer functions. The fused visualization is executable with interactive frame rates.

References [1] Koji Koyamada, Naohisa Sakamoto and Satoshi Tanaka, “A

Particle Modeling for Rendering Irregular Volumes,” Proceedings of the International Conference on Computer Modeling and Simulation (UKSIM 2008), Cambrige, England, April 13, 2008, pp372-377, 2008.

[2] Satoshi Tanaka, Kyoko Hasegawa, Yoshiyuki Shimokubo, Tomonori Kaneko, Takuma Kawamura, Susumu Nakata, Saori Ojima, Naohisa Sakamoto, Hiromi T. Tanaka, and Koji Koyamada, “Particle-Based Transparent Rendering of Implicit Surfaces and its Application to Fused Visualization”, EuroVis 2012, Vienna (Austria), pp.25–29 (short paper), 2012.