VOs may be acquired at two points in time: either preoperatively or intra-operatively. About two decades ago, first IGOS systems were introduced that were based on preoperatively acquired computed tomography (CT) scans. The advantage of this modality is that it provides excellent bone-soft tissue contrast and is geometrically undistorted. These advantages make CTs superior to magnetic resonance imaging (MRI) as preoperative VOs, although the latter method has clear advantages regarding radiation exposure to the patient. Some efforts have been made to overcome the MRI-related difficulties [6, 7]; however, up to now CT remains the method of choice of preoperative imaging for IGOS applications. Another drawback of preoperative VOs led to the introduction of intra-operative imaging modalities. The bony morphology may have changed between the time of image acquisition and the actual surgical procedure. As a consequence, the VO may not necessarily correspond to the TO any more leading to unpredictable inaccuracies during navigation or robotic procedures. This effect can be particularly adverse for traumatology in the presence of unstable fractures. To overcome this problem in the field of surgical navigation, the use of intra-operative CT scanning has been proposed [8], but the infrastructural changes that are required for the realisation of this approach are tremendous, often requiring considerable reconstruction of a hospital’s facilities. An alternative is the usage of established intra-operative imaging modalities. Several research groups developed navigation systems based on fluoroscopic images [9, 10]. The image intensifier is a well-established device during orthopaedic and trauma procedures and could therefore be integrated into IGOS systems easier than intra-operative CT machines. However, the images generated with a fluoroscope are usually distorted, which is caused by a number of factors. To use these images as VOs therefore requires the calibration of the fluoroscope involving the attachment of marker grids to the image intensifier and the tracking of its position and orientation with the navigator during image acquisition [9, 10]. The resulting real-time visual feedback provided by the navigation system (see Fig. 49.2) is similar to the use of the fluoroscope in constant mode. This technique is therefore also known as “virtual fluoroscopy” [11]. Although only two-dimensional (2-D) projections are available and the images usually lack contrast when compared to CT scans, the advantages of fluoroscopy-based navigation preponderate for a number of clinical applications.

In order to address the 2-D projection limitation, a new imaging device was introduced [12] that enables the intra-operative generation of 3-D fluoroscopic image data. It consists of a motorised, iso-centric C-arm that acquires series of 50–100 2-D projections and reconstructs from them 13 × 13 × 13 cm³ volumetric datasets which are comparable to CT scans. Being initially advocated primarily for surgery at the extremities, this “fluoro-CT” has been adopted for usage with a navigation system and has been applied to several anatomical areas already [13–15]. As a major advantage, the device combines the availability of 3-D imaging with the intra-operative data acquisition. “Fluoro-CT” technology is under continuous development involving smaller and noniso-centric C-arms, “closed” C-arm, i.e. O-arm^TM design [16], faster acquisition speeds, larger field of view, and also flat panel technology.

A last category of navigation systems functions without any radiological images as VOs. Instead, the tracking capabilities of the system are used to acquire a graphical representation of the patient’s anatomy by intra-operative digitisation. Using any tracked instrument, the spatial location of anatomical landmarks can be recorded. Combining the obtained points into lines and surfaces will step-by-step generate an abstract model of the geometry. Because this model is generated by the operator, the procedure is known as “surgeon-defined anatomy” (SDA). The technique is particularly useful when soft tissue structures such as ligaments or cartilage boundaries are to be considered that are difficult to identify on CTs or fluoroscopic images. Moreover, with SDA-based systems, some landmarks can be acquired even without the direct access to the anatomy. For instance, the centre of the femoral head, which is an important landmark during total hip and knee replacement, can be reconstructed from a recorded passive rotation of the leg about the acetabulum. It should be noted that the generated images are often rather abstract and not easy to interpret as exemplified in Fig. 49.3. Sati and co-workers suggested the superposition of a preoperative X-ray to facilitate orientation [17], but the precise matching of the two image spaces turned out to be difficult. An alternative concept is provided by the so-called bone morphing [18, 19]. This technology uses a database of generic 3-D statistical computer models of bones and a set of patient-specific points that are acquired with the SDA technique. Analysing the recorded data lets the system select that bone model from the data pool that best matches the patient’s morphology. A special morphing algorithm would then deform the selected model until it fits the acquired points as good as possible [20]. As the result, a realistic virtual model of the operated structure can be presented and used as a VO without any conventional image acquisition (Fig. 49.4).