Radiotherapy planning of tumours in the thorax and upper abdomen is commonly based on the use of a single computed tomography (CT) scan performed during quiet respiration, in contrast to diagnostic CT scans, which are performed during breath-hold. The reason for performing CT scans during quiet respiration is to reproduce the situations during daily treatment that take place under similar conditions. However, this introduces artefacts that incorrectly characterise the geometric shape and extent of a structure.³ As has been demonstrated using phantom studies^,4 the extent of artefacts depends on the interplay between CT slice acquisition and the asynchronous motion of different organs. This may result in tumours being imaged in two or more distinct parts, with the axial slices being shuffled out of order (see Figure 1). Furthermore, nonrepresentative imaging using a conventional planning CT scan also leads to incorrect information on normal organ size and position, with inaccurate information on the actual dose-distribution in these structures. Respiration-induced organ motion greatly degrades the potential effectiveness of advanced treatment planning and delivery. Until recently, the problem of doses to mobile normal organs was largely ignored in routine practice, but standard ‘safety’ margins were added around tumours in order to ensure target coverage. Studies have shown that even these margins may be insufficient to account for extremes of mobility ^5–7 and may increase the risks of toxicity to surrounding organs, which in turn limits the total dose of radiotherapy.

Individualised Approach to Tumour and Organ Mobility
Individualised, i.e. patient-specific, margins are required to account for mobility as no clear correlation exists between mobility and anatomical location in the thorax.6–8 A major advance in this field is respiration-correlated (4-D) CT scanning in which spatial and temporal information on organ mobility are generated using cine scans, while the respiratory waveform is synchronously recorded during imaging. As multiple axial CT slices are acquired at each table position for at least the duration of one full respiratory cycle, this approach yields complete data on organ mobility in the prescribed trajectory. The clinical data described below was mainly generated on a 16-slice CT scanner during quiet uncoached respiration. During the scanning procedure, respiratory signals are recorded using Varian Real-Time Position Management (RPM) respiratory monitoring hardware. The RPM system uses infrared-reflecting markers mounted on a block that is placed on the upper abdomen. Advantage 4-D software was used to sort each CT image into one of 10 ‘bins’ corresponding to the respiratory phase at which the image was captured. A 4-D CT scan is essentially a collection of 3-D CT volumes at all different respiratory phases in the patient, all of which is acquired in a single session. Such a full 4-D scan of the whole thorax is typically acquired within one to two minutes.

Other methods that have been described for acquiring co-registered respiratory signals include spirometry, nasal thermocouples and abdominal strain gauges. If 4-D information is to be used for gated treatment delivery, the method used to acquire respiratory information should also be present at the treatment unit.