Visualization Tools

One area in which 3-D interactive displays are making inroads is minimally invasive or ``key-hole'' surgery [80,11,61]. This surgery relies on inserting (for instance) light through one small hole in the body, a telescope through a second, and surgical tools through a third. By minimizing the amount of tissue cut, this form of surgery reduces the trauma done to the patient during surgery. Minimally invasive surgery was introduced in 1987; by 1992 it was used for about 85% of gallbladder operations performed in the United States [16]. A problem has been that the surgeon is typically provided with only a two-dimensional display, whereas operative procedures take place in three dimensions. At least four companies are competing to provide the third dimension: International Telepresence in Vancouver, Canada, American Surgical in Boston, U.S.A., Storz in Tuttlingen, Germany, and Wolf in Knittlingen, Germany [16]. International Telepresence provides the third dimension by looking through a single lens from different angles; the other three companies use two lenses.

Latent Image [22] has a process for converting 2-D images to 3-D images for tasks such as processing diagnostic images and medical illustration and training. The process involves performing a parallax shift of the image and then ``patching'' missing information, using either other images of the same objects or human intervention. As with synthesizing color from black-and-white pictures, the result is not necessarily accurate, but is useful for some tasks.

The Mayo Clinic has developed COMPASS, a stereotactic computer interactive robotic technology which allows CT and MRI based volumetric tumor resection [34].

Integrated Surgical Systems [56] has been developing a system which combines computer-based medical imaging and CAD/CAM for hip replacement operations using image-guided robotics. In a preliminary phase, the surgeon inserts three locator pins into the patient which act as markers to define a bone coordinate system. CT scans are used for preoperative planning. The surgery is planned by manipulating 3-D images of the femur and the implant. During surgery, the robot uses the locator pins to convert the planned operation into an actual operation.

The University of North Carolina has composited live camera video images with ultrasound images transformed to correspond to the viewer's current position. The effect is an image which appears stationary inside a patient as the observer changes position [1].

Several organizations are building detailed three-dimensional models of the human body. In the short-term, these models fall under the category of visualization tools, because they tend to be too huge for affordable computational engines to support at useful levels of interaction. As computational prices continue to fall, however, it is to be expected that a wide range of simulators will be built using these models.

The United States National Library of Medicine has underway the Visible Human Project [75], which aims to be the most comprehensive digital record of the entire human body ever assembled, stored at millimeter resolution. The project will make use of data from CT, MRI, and photographs of millimeter-thick slices.

The University of Texas Medical Branch at Galveston (UTMB) and Lincom Corporation are developing an anatomical virtual environment which is intended to enable medical students and surgeons to emerse themselves within a specific region of the human body [57]. This is viewed as a first step towards a simulator for complex, high-risk surgical procedures. Currently, the system uses the Marching Cubes algorithm on CT and MRI scans to create a polygonal representation. Since current workstations cannot manipulate these images in real-time, they are stored on laser disks which can be sequenced to provide stereoscopic images.

The Digital Anatomist Program at the University of Washington's Department of Biological Structure has been developing a digital model of the human body for over ten years [76,5,4,17,47,48,60,74]. This project has taken a labor-intensive approach to developing 3-D anatomical models. Beginning with 2D tissue sections, contours are hand-digitized by anatomists, then combined by computer into 3-D shapes which can be displayed as computer graphics. As opposed to ``raw'' CT or MRI pictures, the result is that the computer has a detailed knowledge of where the anatomically significant boundaries are. This process has been colorfully described by the departmental chair as ``teaching a computer anatomy''.

The Vesalius Project at Colorado State University [46] is similar to the Digital Anatomist Program.

Stanford's Electric Cadaver [85] is a Hypercard program on the Apple Macintosh that uses the Bassett slide collection of photographs of anatomic dissections, plus line drawings as the main image source for learning about anatomy. It does not contain 3D images or animations.