Synergistic imaging

The term synergistic imaging is used to describe the successful combination of information from more than one medical imaging modality, a process which provides more useful clinical information than the individual component images. This process is becoming an increasingly active area of research with important application to nuclear medicine. Several factors have contributed to making synergistic imaging feasible in the clinical environment. Increasingly effective interfaces to image acquisition devices have been developed, and the increased performance coupled with the decreased cost of computer workstations necessary for the sophisticated image manipulation have facilitated the introduction of digital networks within hospitals, thus creating the infrastructure for clinical application. Synergistic imaging may well prove to be the major advance in clinical imaging in the 1990s. Already, it has been used successfully in planning stereotaxic neurosurgery and, with the aid of registered high resolution images, improves interpretation and quantitation in nuclear medicine studies, particularly of the head. It is likely that the quality of imaging of other regions of the body, especially the heart, pelvis, spine and abdomen would be enhanced by a multi-modal approach. Whenever two or more imaging modalities are used in clinical investigations, opportunities for synergism exist. This editorial focuses on applications of synergistic imaging which use nuclear medicine images. The main applications in nuclear medicine can be categorised as follows: 1. To assist in image interpretation by improving anatomical localisation (Hawkes et al. 1990, 1991b; Pelizzari et al. 1989; Schad et al. 1987). 2. To improve the accuracy of quantitative studies by assisting in delineation of regions or volumes of interest around anatomical structures in order to relate functional information such as tissue perfusion and metabolism to structure (Evans et al. 1988, I991; Mazziotta et al. 1991 ; Valentino et al. 1988). 3. To combine or "fuse" information contained in each imaging modality to produce a new representation. Information contained in each image is used in a quantitative way. Examples include better calculation of attenuation correction in single photon emission tomography (SPET) (Fleming 1989) or improving the modelling of the partial volume effect in positron emission tomography (PET). In the future it may well be possible to relate information on tissue perfusion derived from nuclear medicine images to information on vascular supply from angiography. Synergistic imaging, regardless of the application, involves three distinct processing stages: to establish very accurate registration, data fusion of these data sets and finally the display of the combined information as a new image. The research and development emphasis to date has mainly focused on solving registration problems. Image registration is conceptually straightforward but can be difficult in practice. The geometric transformation relating one image to another is derived from the image co-ordinates of corresponding points or features. Two main methods of registration have emerged, one using anatomical features visible in the different modalities, the other using external markers. The external markers are fixed to the patient and the image co-ordinates of these markers are used to derive the transformation of the images. Markers attached to the skin surface of the head are frequently used although skin movement limits their accuracy to 3~4 ram. They are difficult to relocate and therefore are usually kept in place for the duration of both sets of scans. It is straightforward to design markers visible with two or more modalities. In our laboratory, we have designed a simple aluminium holder for a cobalt-57 source for registering the radiograph and planar isotope bone scan in the management of trauma of the wrist (Hawkes et al. 1991 b). For registering magnetic resonance (MR) and SPET images of the head in three-dimensions (3D), we use markers consisting of a spherical void moulded in silicon rubber and attached to the skin with electrocardiography (ECG) electrode tape. The spherical void is filled with a fluid which is visible in each modality (Hawkes et al. 1990). The diameter of the void is larger than the slice spacing of the most coarsely sampled modality. As an alternative system, the Utrecht group have reported the use of " V " shaped markers consisting of a small triangular shaped slab visible in CT, to which small tubes are attached to two sides (Van den Elsen and Viergever 1991). The tubes contain a fluid which is visible in SPET and MRI. The marker co-ordinate is defined as the location of the tip of the " V " or triangle. This marker design is capable of accurate determina-