Grid to fight cancer
Radiotherapy Grid computing offers an application of high-energy physics research-work- still-in-progress with tangible and immediate clinical utility to the one million Americans who will be treated with radiotherapy next year. The “voices” article from the first issue of symmetry does promise a deep and wide resonance indeed.
In a computer simulation, pencil beams of ionizing radiation are directed towards a finely segmented calorimeter. The challenge: determine the optimal aiming of these beams so that a specified amount of energy is deposited to certain target volumes of the calorimeter, while non-target volumes are avoided. For bonus points: run enough particles in the simulation so it results in a sharply defined energy deposition per voxel (volume element).
This is not a proposal to test new particle physics calorimeters, it is cutting edge radiotherapy treatment planning. Patients indicated for radiotherapy are imaged prior to their treatments, usually via computerized tomography. Typical imaging voxel segmentation is a few millimeters.
Cancerous lesions are designated as target-volumes, while nearby uninvolved organs are characterized as targets-to-be-missed. This is a straightforward calculation. Alas, the computer running time needed under realistic clinical conditions, even on a modest size parallel cluster, exceeds a workday. To increase calculation speed, the vast majority of radiotherapy planning is done today with algorithms of intrinsically reduced accuracy.
Dimitri A. Dimitroyannis
Harvard Medical School, USA