Dr Virgiliu Craciun leads our radiotherapy physics research as well as acting as an MPE in external beam planning.
Current areas of research interest
Computed radiography for thoracic dosimetry in total body irradiation
Total body irradiation (TBI) is used to prepare for bone marrow transplant and in treating haematological malignancies. Radiation induced interstitial pneumonia can be fatal, so the dose the lungs receive is critical, and needs to be carefully measured.
This project aims to develop an efficient dosimetric method for screening the MV radiation dose in lungs during TBI treatment. It uses a computed radiography (CR) cassette designed to be used with the higher energy beams and doses used in radiotherapy. Calibrating the pixel values in a CR image for a dose would produce a 2D map of the dose distribution. It is a quicker, automated system that would enable dosimetry to be performed for all treatment fractions. A digital dose distribution could be visualised in various ways such as a colour-maps, varying greyscale intensity and isodose lines, and data analysis will be fast.
The pictures below show our treatment system with the CR in place (left) and a digital dose colour map of the thorax and superior abdomen (right).
Optimised algorithms for accelerated dose computation on a graphics process unit (GPU) four-dimensional vector space
We're testing whether we can calculate the whole dose distribution in the patient using the OpenGL graphics industry open standard on GPU, instead of the current proprietary GPU methods, like CUDA. We are interested in developing highly optimised algorithms that will use the four-dimension GPU colour vector space to perform parallel scalar radiation dose computation. This would mean faster performance and wider applications. We use C/C++, GLSL and scripting languages to implement the algorithms on multi-core CPU (central processor unit) and GPU.
Semi-automatic validation of radiotherapy treatment plans
The aims of this project are to:
- develop prototype software to perform validation quasi-automatically
- facilitate a higher level of real-time treatment plan analysis than is currently performed manually
- significantly reduce the required time and skills.
The validation stage for radiotherapy treatment plans involves checking the whole anatomy outlined as geometrical structures, checking that the radiation dose distribution in tumours and normal tissue is correctly devised, calculated and conforms to protocol, and also if any improvement is recommended.
We're working on elaborating algorithms to assess the correctness of geometrically represented organs in correlation with medical images, and algorithms for analysing 3D radiation dose grids with respect to beam parameters, determining if further optimisation is possible.
Heuristic analysis of anatomical structures using spike detection
Following initial collaboration with the University of Southampton, under a FPAS-SoM grant we are working to devise an algorithm that can analyse in an heuristic mode the possible flaws that may occur in geometrically delineated anatomical structure. While the current method of scrutinising each contour, on each slice for each anatomical structure, needs human input and can be time consuming, this heuristic algorithm will automatically analyse the contoured geometries and, based on the general principles of the human organs, will try to offer hints of possible geometrical errors. This is built around a spike detection algorithm used in protein spectrometry, but transformed and adapted to work with the characteristics of the human structure.
We also provide scientific support to the set-up and recruitment to many radiotherapy clinical trials. This work is led by Charlotte Britton.