Clinical Aspects of Image Guidance and Localization in Radiotherapy

Workflow Example 1: CT-Based IGRT

A first step in IGRT and a key foundation is the reference image used for treatment planning (in this case the simulation CT, Figure 18.1A ). An isocenter and region-of-interest structure sets are defined in the treatment planning system. These are transferred (if necessary) to the image guidance system. When the patient is ready for treatment and positioned at the intended isocenter, a CT is acquired ( Figure 18.1B ). This CT is aligned with the reference data set using an interactive image fusion software interface ( Figure 18.1C ). Registration uses rigid body alignment with 3D translations and (optionally) rotations around three axes. Registration can be accomplished manually or automatically via various algorithms that can be set to focus on particular regions of the image (e.g., a 3D box) or on different threshold values of the CT number (e.g., bony features or soft tissue gray values). The registration results in a calculated shift correction. The patient is moved according to these shift calculations, a process that is often accomplished automatically via a command to the linac table. Some patient support systems (e.g., HexaPod) allow for “six-degree-of-freedom” positioning (i.e., translational movements in three directions as well as rotations of up to 3 degrees about the three principle axes). One issue is that the rotation may correctly describe the anatomic alignment near the isocenter but may be less accurate away from the isocenter. A rotation of 1 degree corresponds to a shift of approximately 2 mm at 10 cm from the isocenter. Such inaccurate alignments off-axis can have implications for doses to organs at risk.

Whatever the correction strategy, it is valuable to have a predefined protocol for IGRT that includes the roles and responsibilities for each professional group and the associated protocols (see ASTRO safety white paper on IGRT ). An example protocol from a fictional hospital is presented in Table 18.2 . The protocol spells out who is responsible for IGRT alignment (therapists in this case). It also lists threshold limits for shifts (10 mm in this case) after which they must consult a radiation oncologist. Such thresholds are a specific recommendation of the ACR/ASTRO practice guidelines for IGRT and ideally should be site-specific and based on an analysis of the random and systematic errors. The example protocol also spells out professional roles. Here the physician and physicist are expected to be present for the alignment procedure on the first day but not thereafter. This would vary for other treatment protocols. In SBRT cases, for example, the physician would typically be present for each treatment.

To this idealized IGRT process are added the many issues and challenges that can arise in actual practice. One well-worn error pathway is related to isocenter placement. The image guidance system must have the correct reference point defined, typically the isocenter from the treatment planning computer. If an incorrect reference point is transferred the patient alignment will be incorrect. This error is particularly insidious because it is not visible to the operator. Similarly, on a TomoTherapy system the reference or red lasers must be set to the reference point in the planning system. Not doing so will lead to systematic errors.

Other, less gross errors are possible. For example, the bladder filling for a pelvis case might be different at the time of treatment compared to simulation. In such a situation it would not be possible to align all the pelvic anatomy without intervening with the patient and rescanning. Figure 18.1B shows an example from a patient with head and neck cancer where part of the anatomy is well aligned (cervical vertebrae) but another part is not (mandible and skull). Such region misalignments represent a significant challenge where extensive regions are being treated and underscore the importance of the physician being involved in defining the alignment objectives for IGRT cases.

Another question is how well the CT planning scan (acquired prior to treatment) represents the anatomy during the actual treatment. Several issues come into play here: patient movement, nonideal immobilization, long wait times between scanning and treatment, changes in tumor size or position during the course of therapy, and changes in patient body habitus (e.g., weight loss). A few publications have addressed this issue, but there is relatively little peer-reviewed literature given the wealth of IGRT data that have been collected. In the future more reports will hopefully appear, providing practical guidance on IGRT setup issues. To evaluate movement on individual cases it may be useful to perform a repeat CT just after treatment to assess movement (perhaps for several fractions). For longer treatments (e.g., SRS/SBRT) repeat imaging may be performed mid-treatment.