Occlusion-based Direct Volume Rendering for Computed Tomography Image

Abstract

Direct volume rendering (DVR) is an important 3D visualization method for medical images as it depicts the full volumetric data. However, because DVR renders the whole volume, regions of interests (ROIs) such as a tumor that are embedded within the volume maybe occluded from view. Thus, conventional 2D cross-sectional views are still widely used, while the advantages of the DVR are often neglected. In this study, we propose a new visualization algorithm where we augment the 2D slice of interest (SOI) from an image volume with volumetric information derived from the DVR of the same volume. Our occlusion-based DVR augmentation for SOI (ODAS) uses the occlusion information derived from the voxels in front of the SOI to calculate a depth parameter that controls the amount of DVR visibility which is used to provide 3D spatial cues while not impairing the visibility of the SOI. We outline the capabilities of our ODAS and through a variety of computer tomography (CT) medical image examples, compare it to a conventional fusion of the SOI and the clipped DVR.

Fig. 1. Conventional direct volume rendering (DVR) of a patient study with a lung carcinoma in the left upper lobe of the lung. A TFdesigned to reveal the lung carcinoma is applied. (a) shows a quadrant view of a multi-planar reformatting (MPR) together with a DVR(bottom-right); (b) shows a DVR fused with a slice of interest (SOI) set to the same coronal view as in the MPR; (c) shows our proposedocclusion-based DVR augmentation for SOI (ODAS) which augments the 3D spatial cues from the DVR while preserving the visibilityof the SOI.

Fig. 2. An overview of the processes involved in ODAS algorithm using a chest CT volume as an example.

Fig. 3. A depiction of the movement of the random view camera,V, to the center of the coronal view, V’, where all structures of theSOI is visible.

Fig. 4. Occlusion-driven dynamic opacity weight curve.

Fig. 5. GPU computation of the occlusion depth map and itsresulting histogram.

Fig. 6. Comparative results between ODAS and DVR techniques. (a) shows a selected SOI, in the coronal view, of a CT image volumedepicting vascular structures and their related organ structures. (b) is the rendering of ODAS with automatically derived depth D = 47,and (c) is the DVR counterpart with volume clipping set to the same D. (d) and (e) are variations of the DVR set to clipping depth of ¡¾20from (c).

Fig. 7. An example of the ability of ODAS to preserve the level ofvisibilities for structures that are close to the SOI when comparedto volume clipping using the same D in MPR (axial and sagittalviews, respectively of left and middle columns) as well as anarbitrary view (right column). Note the different D calculationsbased on the view-point and arrows indicating the missing vesselsthat are visible with the ODAS.

Fig. 8. Results from using different percentile values on theocclusion depth histogram.

Fig. 9. Slice-by-slice volume navigation with ODAS in sagittalview.

Fig. 10. Applications of TF to the ODAS with 1D TF in (b) and2D TF in (c). In both of these renderings, the TFs were designedto highlight the lung nodule structure.

Fig. 11. Averaged computation time for individual processes inour ODAS algorithm.