Modern medical technology provides a wide range of scanning and measurement systems. Radiologists and surgeons often have to interpret a large number of images from different types of scanners. A visualization application based on Visualization Data Explorer* (DX) allows users to obtain qualitative information (shape, relative position) and quantitative data (distance, volume, area) on selected 2D/3D objects. This information can be used for diagnosis and subsequent treatment planning. With DX , the user can create opaque and transparent renderings of 3D regions. With interactive rotate, zoom, and pan, the user can easily study key objects. Using these techniques, radiologists could fully utilize the results of costly scans and minimize the time required to obtain relevant clinical information.
The following two examples demonstrate the effectiveness and benefits of Data Explorer . The first example is based on routine 3D SPECT (Single Photon Emission Computed Tomography) and CT (Computed Tomography) scans of a patient with liver lesions. 3D SPECT is an effective tool, enabling the radiologist to visualize where radioactivity has accumulated. However, it poses two special visualization problems: low resolution and diffuse boundaries. Identification of specific organs or tissues, as well as their exact shape and features is difficult at best. The CT scan provides important referential information when correctly aligned with the SPECT data. The second set of examples are made from MRI (Magnetic Resonance Imaging) angiogram scans of the head and neck.
Typically, thoracic studies are designed to determine the position and size of lesions. Several of Data Explorer's 2D and 3D tools support interactive analysis and display of SPECT data. When the 3D CT data set is registered with the SPECT data, CT images and objects can also be displayed with the 3D SPECT data.
To clearly show the regions of high and low radiopharmaceutical accumulation, a surface surrounding the high activity is rendered opaque, and a surface representing the boundary of the low activity region is rendered transparent. (See Figure 1).
The opaque surface (high activity) corresponds to the localized abnormal accumulation of radioactivity within the organ. The SPECT scan is based on radiopharmaceuticals that are tumor-avid. The transparent surface corresponds to the surface of the involved organ. The radiologist can display contours on the opaque and/or transparent surfaces to improve shape perception and estimate the size of various regions. In Figure 1, axial contours spaced 6.0 mm apart are placed on the opaque surface. The volume enclosed by the red surfaces is computed.
Before displaying the CT scan with the SPECT data, the radiologist registers coordinate frames. The coordinates of the SPECT data are translated and rotated relative to the CT coordinate frame to ensure that a given 3D coordinate refers to the same anatomic location in the SPECT and CT data sets. Two approaches are available for registration: direct 3D registration based on interactive alignment of surfaces, and iterative registration based on 2D images and 3D surfaces.
When the SPECT and CT data sets are registered, planar images can be displayed with 3D SPECT surfaces. This can be a powerful combination as 2D images provide detailed context information in the plane, and the 3D surfaces shows the overall shape and size. (See Figure 2)
With a registered CT data set, 3D SPECT surfaces can be displayed with anatomic objects extracted from the 3D CT data. An example is shown in Figure 3. The ribs, spine, stomach, left kidney, and aorta are clearly seen. (Note the 3D reconstruction is based on only 13 2D image slices from the CT data set.) Using Data Explorer, the user can extract additional quantitative details from this composite image.
MRI angiographic scan sequences produce a high contrast between vessels and tissue. Figure 4 shows how clearly the vessels of the brain can be visualized.
The image shown above was built from a data set of 120 2D MR images, each 512x512. By forming this surface we've reduced the volume of data from 63 MB to 4.9 MB. As a result, this surface is small enough that the radiologist can rapidly rotate and zoom the image to explore whatever is of clinical interest. By using Data Explorer's "pick" module, the user can measure the lengths of vessel segments and distances between key structures, as shown in Figure 4. This module also allows the user to select a vessel segment, obtain the volume enclosed by the vessel segment, and measure the cross sectional areas. In Figure 5, the user is measuring the area of a region of the left carotid artery. These measurement tools provide a direct means of characterizing vessel restrictions and aneurysms.
Medical and technical assistance were generously given by Prof. Marilyn Noz and Dr. Elissa Kramer at the New York University Medical Center, Department of Radiology.
[ IBM home page | Order | Search | Contact IBM | Help | (C) | (TM) ]
Figure 1 (36 KB image)
Figure 2 (45 KB image)
Figure 3 (58 KB image)
Figure 4 (25 KB image)
Figure 5 (23 KB image)