Dual energy image reconstruction and systems for application in proton therapy treatment planning

Abstract

Proton therapy is the use of a proton beam rather than a traditional X-ray beam in the treatment of cancer. This technique is being developed all over the world due to the unique Bragg peak feature of proton beams. In order to guarantee accurate dose delivery to the tumour, the stopping power ratio (SPR) of the tissue must be known. This parameter is dependent on the electron density and effective atomic number of the material and describes the energy loss per unit length in the tissue. In current clinical practice, the SPR of patient tissues is obtained through single energy CT (SECT) scanning. The SECT scan results in a map of kilovoltage X-ray attenuation coefficients relative to the attenuation coefficient of water for the beam energy. This quantitative information is then converted to SPR via an empirically derived look-up table. If the patient tissues do not have a similar chemical composition to the materials used to generated the look-up table, this approach can lead to diminished dose calculation accuracy. As a result, the patient may experience increased normal tissue complication or decreased tumour control probability. An alternative approach that has been suggested recently is the use of dual energy CT (DECT). DECT is an emerging imaging modality that makes use of CT spectra to create two sets of CT images simultaneously. DECT relies on the energy independence of relative electron density, and the energy and atomic number dependence of X-ray interaction atomic cross-sections. Post processing of the two reconstructed CT images results in two separate images quantifying electron density and effective atomic number. The SPR of the tissue can be calculated once electron density and effective atomic number are known. In theory, the use of DECT for SPR estimation should be more robust than SECT combined with an empirically derived look-up table. This hypothesis has been tested with phantoms of known composition in the current work. Unfortunately, the post processing of DECT images results in effective atomic number images with a low contrast to noise ratio, which can affect SPR calculation accuracy. To counteract this, an iterative DECT image reconstruction approach has been developed. Two image reconstruction algorithms, FBP and TVS-DROP, are implemented to reconstruct the CT images, where an advanced parallel calculation code was designed for TVS-DROP to improve work efficiency. The iterative reconstruction algorithm was also applied to a radioisotope form of cone beam DECT. A feasibility study into the use of this novel imaging method in adaptive proton therapy was conducted. In summary, the objective of this thesis is to examine the application of DECT for proton therapy treatment planning, develop improved image reconstruction techniques for DECT, and investigate the feasibility of a novel radioisotope-based form of cone beam DECT for adaptive proton therapy.