Radiotherapy is the use of ionizing radiation to treat cancer. Sophisticated software programs known as treatment planning systems (TPS) are commercially available to tailor-make treatment plans for individual patients. These systems mainly allow clinicians to predict the dose delivered to each voxel in the patient for a given orientation of the radiation beams and other parameters used before actually treating the patients. In modern TPS the fluence of the each beam can be modulated spatially and temporally (aka, IMRT and VMAT) in order to spare normal tissues as much as possible without compromising the dose to the tumor. Several dose computation models exists ranging from simple analytical methods to most accurate and time consuming Montecarlo methods to predict the dose delivered to each voxel within the patients’ body. The key component of the dose computation model is the radiological path length calculation which accounts for tissue heterogeneities that affect the dose delivered to different types of tissue.
Ray tracing is widely used in computer graphics for virtual reality rendering. The same concept is used in radiotherapy applications for calculating the exact radiological path length (RPL). A patient 3D model is generally obtained from a set of greyscale 2D CT slices. The greyscale values which represent Hounsfield units (HU) of the voxels can be used to calculate the RPL by using the relationship between HU and relative electron density (RED). Prior to dose calculation the RPL has to be calculated for each beam and this is a per-voxel operation, i.e., the calculation has to be done for each voxel and this is time consuming. As the number of beams increases the calculation time increases substantially.
A ray tracer module has been developed in C++ and incorporated into an in-house developed software with the front end developed using Qt. This software serves as a research platform to test new algorithms for use in radiotherapy.
The aim of this project is to use parallel processing libraries such as OpenMP to parallelize the task in order to speed up the RPL calculation.
This would suit a master student with a knowledge of parallel processing and C++ programming
C++ programming, computer vision, decision support systems. mathematical modelling, medical informatics, parallel processing, signal processing