A compact multileaf collimator for conventional and intensity modulated fast neutron therapy

A 120 leaf collimator of high resolution has been constructed to shape a fast neutron therapy (FNT) beam produced from a superconducting cyclotron using the d(48.5)+Be reaction. The computer controlled multileaf collimator (MLC) replaces an aging, manually operated multirod collimator (MRC). The MLC was built to address two problems: The need to increase the efficiency of FNT at the Gershenson Radiation Oncology Center of the Karmanos Cancer Institute at Detroit, MI, and the desire to implement intensity modulated neutron radiotherapy for which a suitable computer controlled beam shaping device of high resolution and rapid shape changing not currently exist. The specific aims were to build a neutron MLC that would solve these problems and then verify its radiological performance as being clinically acceptable. The MLC leaves project 5 mm in the iso-centric plane perpendicular to the beam axis. A taper has been included on the leaves matching the beam divergence along one axis. A 5 mm leaf projection width was chosen to give high resolution conformality across the entire field. The maximum field size provided is 30x30 cm2. To reduce the interleaf transmission a 0.254 mm blocking step has been included. End-leaf steps totaling 0.762 mm were also included allowing adjacent leaf pairs to close off within the primary radiation beam. The neutron MLC also includes individual 45 deg. and 60 deg. automated universal tungsten wedges. All electro-mechanical functions of the MLC are governed by the multileaf collimator control system (MLCCS). The MLCCS control functions are leaf and wedge motion control, leaf and wedge position verification, switching on and off the radiation beam, and inputting certain cyclotron interlocks statuses as well as outputting MLC interlocks. The MLCCS is responsible for setting required shapes on the MLC, and verifying those shapes to be accurate to within ±1 mm for each leaf's projection in the plane perpendicular to the beam axis at the position of the iso-center. The position verification function uses a machine vision system that images optical targets on the leaves to verify set field shape accuracy within ±1 mm at the level of iso-center. The MLC transmission was measured to be 3.9±0.5% with an 11±2% gamma component which is slightly lower than the MRC measured transmission of 4.6±0.5% with a 17±2% gamma component. The difference in the measured gamma component ratio for the closed collimators agrees within 40% to a theoretical approximation of ∼2.4 for tungsten to steel when considering equal attenuation lengths. The actual MLC penumbral measurements in water for a 10x10 cm2 field at 5 cm depth were 9.1±0.2 mm and 11.8±0.5 mm along the focused and unfocused axes, respectively. The MRC penumbra measured 8.6±0.3 mm under the same conditions along both of its focused axes suggesting penumbral equivalence between the MLC and MRC along the focused axis with slight degradation in the unfocused direction of the MLC. The many benefits of the fully automatic MLC over the semi-manual MRC are considered to justify this compromise