Immobilization Techniques and Error analysis

Almon Shiu Ph.D., Congjun Wang, Ph.D., Catherine H. Wang, Ph.D.,

Lifei Zhang, Ph.D., Lei Dong, Ph.D.

Radiation Physics Department, The University of Texas M.D. Anderson Cancer Center

Houston, TX USA

Patient setup uncertainty is one of the major concerns in radiation therapy, especially in high precision treatments such as 3D conformal and IMRT (intensity modulated radiotherapy) techniques. The setup uncertainties of various immobilization techniques for brain, head and neck, and spinal tumors were evaluated. Either the 2D to 2D or the 3D to 3D image registration methods was used to assess the setup accuracy. For the 2D to 2D method, AP and RT LAT digital Portal images were acquired and registered with the AP and RT LAT digital reconstructed radiography (DRRs) from the axial Planning computed tomography (CT) images. The METEC Plastic mask system and the Gill-Thomas-Cosman (GTC) non-invasive head frame system, used for brain lesion treatments, were evaluated with the 2D to 2D method. For the 3D to 3D method, in-room CT scans were acquired using a CT-on-rails system for 14 head and neck (H&N) patients, 2 acoustic neuronal patients (56 treatments), and 15 spinal and para-spinal patients with Varian Exact couch and 20 spinal and paraspinal patients with 6D robotic couch top for the treatments. Two SRT patients with a non-invasive head frame (GTC frame) were treated using image-guided SRS technique for ten treatments. The previous SRT setup reproducibility from over 250 treatments is within 1 mm with the aid of the five-point GTC alignment device. This alignment device is minimizing the misalignment of the GTC frame in AP, lateral, and axial translations and the angular deviations about each of these axes, which are roll, tilt, and spin. Prior to each treatment, a daily CT scan with a 9-rod localization device was acquired. Daily CT images were registered with planning CT images and/or MRI images. Then, we identified the 9-rod on the daily CT images, such that all the target, critical structures, and external contours from the planning images were transferred to the daily stereotactic coordinate system. Finally optimized the dose distribution with the patient’s daily setup. For H&N patients, three separate bony ROIs were defined: C2 and C6 vertebral bodies and the palatine process of the maxilla. Translation shifts of 3 ROIs were calculated relative to the marked isocenter on the immobilization mask. For the spinal patients, Pretreatment CT scans were fused with the planning CT scans to determine the correct target isocenter by accounting for any translational and roll (axial) rotational discrepancies from the planning CT. (Corrections caused by yaw and pitch rotations have not yet been implemented.) The reproducibility of the treatment isocenter as compared with the planned isocenter was measured with digitally reconstructed radiographs (DRRs), portal film imaging, and immediate post-treatment verification CT scans (the first 30 patients). For the ten treatments of two SRT patients, the mean deviation in AP, LAT and VERT direction is 0.2±0.12, -0.4±0.24, and -0.1±0.07 mm, respectively. For the H&N patients, the shifts for all 3 ROIs were highly correlated. However, noticeable differences on the order of 2-6 mm existed between any 2 ROIs, indicating the flexibility and/or rotational effect in the H&N region. The palatine process of the maxima had the smallest right-left shifts because of the tight lateral fit in the face mask, but the largest superior-inferior movement because of in-plane rotation and variations in jaw positions. The neck region (C6) had the largest right-left shifts. There was no statistically significant improvement for using the S-board (8 out of 14 patients) vs. the short face mask. On the basis of the pretreatment CT scans of the spinal patients, the daily setup deviation from the planned isocenter in each direction was determined for 90 treatments. The deviation from the planned isocenter ranged from –6.0 mm to 4.9 mm in the lateral direction, from –5.5 mm to 5.0 mm in the AP direction, and from –8.0 mm to 4.5 mm in the superior-inferior (SI) direction. Mean deviations and standard deviations (mean deviation ± standard deviation [± SD]) were 0.19 ± 2.69 mm in the lateral direction, -0.4 ± 1.8 mm in the AP direction, and –1.58 ± 3.09 mm in the SI direction. The results clearly indicated that, using the stereotactic body-frame alone, is not possible to achieve the setup accuracy necessary for delivering a highly conformal dose of radiation to the target. With almost real-time CT image guidance, the patient is aligned with the corrected daily isocenter. The deviation from the planned isocenter in each of the three directions was within 0.5 mm (mean deviation ± SD: 0.05 ± 0.36 mm in the lateral direction, 0.08 ± 0.33 mm in the AP direction, and -0.02 ± 0.47 mm in the SI direction). The results for the 20 patients with 6D robotic table-top are also presented. The variability in setup corrections for the different immobilization techniques should be aware when the treatment margins are used in the design of the treatment plan.