Quality assurance of serial tomotherapy for head and neck patient treatments

Abstract

Purpose: A commercial serial tomotherapy intensity-modulated radiation therapy (IMRT) treatment planning (Peacock, NOMOS Corp., Sewickley, PA) and delivery system is in clinical use. The dose distributions are highly conformal, with large dose gradients often surrounding critical structures, and require accurate localization and dose delivery. Accelerator and patient-specific quality assurance (QA) procedures have been developed that address the localization, normalization, and delivery of the IMRT dose distributions.

Methods and Materials: The dose distribution delivered by serial tomotherapy is highly sensitive to the accuracy of the longitudinal couch motion. There is also an unknown sensitivity of the dose distribution on the dynamic mutlileaf collimator alignment. QA procedures were implemented that assess these geometric parameters. Evaluations of patient positioning accuracy and stability were conducted by exposing portal films before (single exposure) and after (single or double exposure) treatments. The films were acquired with sequential exposures using the largest available fixed multileaf portal (3.36 × 20 cm²). Comparison was made against digitally reconstructed radiographs generated using independent software and appropriate beam geometries. The delivered dose was verified using homogeneous cubic phantoms. Radiographic film was used to determine the localization accuracy of the delivered isodose distributions, and ionization chambers and thermoluminescent dosimetry (TLD) chips were used to verify absolute dose at selected points. Ionization chamber measurements were confined to the target dose regions and TLD measurements were obtained throughout the irradiated volumes. Because many more TLD measurements were made, a statistical evaluation of the measured-to–calculated dose ratio was possible.

Results: The accelerator QA techniques provided adequate monitoring of the geometric patient movement and dynamic multileaf collimator alignment and positional stability. The absolute delivered dose as measured with the ionization chamber varied from 0.94 to 0.98. Based on these measurements, the delivered monitor units for both subsequent QA measurements and patient treatments were adjusted by the ratio of measured to calculated dose. TLD measurements showed agreement, on average, with the ionization chamber measurements. The distribution of TLD measurements in the high-dose regions indicated that measured doses agreed within 4.2% standard deviation of the calculated doses. In the low-dose regions, the measured doses were on average 5% greater than the calculated doses, due to a lack of leakage dose in the dose calculation algorithm.

Conclusions The QA system provided adequate determination of the geometric and dosimetric quantities involved in the use of IMRT for the head and neck. Ionization chamber and TLD measurements provided accurate determination of the absolute delivered dose throughout target volumes and critical structures, and radiographic film yielded precise dose distribution localization verification. Portal film acquisition and subsequent portal film analysis using 3.36 × 20 cm² portals proved useful in the evaluation of patient immobilization quality. Adequate bony landmarks were imaged when carefully selected portals were used.

Introduction

A commercial serial tomotherapy planning and delivery system¹was recently installed and commissioned in our clinic (1). The system is used to deliver conformal dose distributions using serial tomotherapy, and includes a treatment-planning system and computer-driven tertiary dy-namic multileaf collimator (DMLC)². The DMLC consists of 40 leaves, arranged in two parallel adjacent banks of 20 leaves each, and is oriented so that a line connecting the projected leaf centers is parallel to the gantry rotation plane. The leaves are pneumatically driven and move in a direction normal to the gantry rotation plane, each projecting on our accelerator to a field size (bixel) of 0.84 × 1.0 cm². The beam is modulated by opening and closing leaves during arc delivery so that, over a narrow gantry angle range (in our case, 5°), the fraction of fluence required from a bixel is equal to the fraction of gantry angle that the leaf is open. The dose calculation algorithm has been described by Wu et al. (2) and Low et al. (3).

Quality assurance (QA) procedures for treatment planning and delivery of traditional conformal therapy have been previously described 4, 5. However, there are significant differences between conventional three-dimensional (3D) conformal therapy and serial tomotherapy treatment deliveries. In conventional therapy, the radiation dose is delivered using portals with high fluence gradients only at portal boundaries. The influence of spatial uncertainties, such as patient movement, can be accounted for by adding a margin to the known and suspected tumor to ensure that the tumor will receive an adequate dose 6, 7. For serial tomotherapy, the radiation dose is delivered by abutting sequential treated volumes, and patient positioning errors of as little as 1 mm can cause dose errors of 10% within the target (8).

The QA of serial tomotherapy dose distributions is more complex than for traditional external beam therapy. For traditional therapy, patient treatments are generated by a set of fixed portals, differing by well-defined degrees of freedom, such as entry angle, portal shape, and intensity. QA of dose calculation and delivery is conducted by definition and verification of a set of experiments encompassing the degrees of freedom encountered in clinical practice (4). The smooth behavior of fixed portal dosimetry as a function of these degrees of freedom allows the definition of a sparse set of test plans to serve as an indication of dose calculation and delivery accuracy for most clinical cases. These dose distributions can be verified using common dosimetry tools and techniques. After a representative selection of beams has been verified, the QA dedicated to individual patients will normally consist of verification of portal geometry, collimator, couch, and gantry angles, source-to–surface distances, and beam modification devices, such as compensating filters. Most of these checks can be conducted by the radiation therapist delivering the treatment and require little added time or effort. The use of record-and–verify systems aids in the efficiency and thoroughness of treatment-specific QA.

Unlike traditional therapy, with serial tomotherapy, no treatment geometry subsets can be used to verify the dose calculation and treatment delivery for a wide range of treatments. Although the dose distribution of an individual bixel can be accurately verified, the large number of bixels used in arc-based serial tomotherapy limits the utility of simple geometric patient-specific verifications. Therefore, a different approach is needed (9).

The accuracy of the intensity-modulated radiation therapy (IMRT) dose distribution calculation algorithm and treatment delivery system was evaluated recently by Low et al. (10), using cubic phantoms (11) and geometric target volumes and critical structures. Ionization chambers, film, and thermoluminescent dosimeters (TLDs) were used. Ionization chamber measurements were used to compare to single-point dose calculations and were found to agree within a standard deviation of 1.0% for regions within the high-dose levels. Ionization chamber measurements also showed an agreement of 1.2% in low-dose regions, averaging 30% of the maximum delivered dose. Spatial localization of the dose distributions was evaluated by using radiographic film placed in the transverse and coronal orientations. The measured and calculated high-dose gradient regions were compared to determine dose distribution localization accuracy. With one exception, the calculated and measured dose distributions agreed within 2 mm.

Verellen et al. (9) described the clinical efficacy of the same tomotherapy system. Patient immobilization and positioning were evaluated using an initial image obtained with a diagnostic x-ray beam and subsequent portal films obtained using the DMLC with all leaves open. Although intertreatment motion was evaluated, no films were obtained to monitor intratreatment motion. Dosimetric verification was conducted using test treatment plans on anthropomorphic phantoms, with spatial positioning of the dose distribution determined using radiographic film and the dose accuracy verified using alanine detectors. Results indicated excellent agreement (discrepancies within 1%) between calculated and measured absolute doses, but doses were measured only for target regions. Doses outside the target were estimated based on TLD measurements placed on the surface of a single patient.

Previous unpublished studies of the dosimetric accuracy of the Peacock treatment planning and delivery system have also been conducted by Grant et al. 12, 13, 14 and Butler et al. (15). We describe the process used to verify the delivered dose to 9 head and neck patients. Additional system QA procedures were implemented to verify sensitive geometric parameters. Patient positioning was verified by comparison of portal films against independently generated digitally reconstructed radiographs. Dose distributions were evaluated for each patient using homogeneous cubic phantoms, with radiographic films used to identify the dose distribution localization, and ionization chambers and TLDs for dose measurements. Doses were evaluated both in high- and low-dose regions. Emphasis was placed on the independent determination of the spatial localization of the measured and calculated dose distributions.