Research Activities

CCMB is internationally known for it’s work in this area.  We have developed physics-based algorithms that are able to accurately predict dose deposited and signal formation in electronic portal imaging systems. Furthermore, we can analyse on-treatment transmission images to reconstruct the three-dimensional radiation dose actually delivered to the patient during treatment.  This allows robust verification of the delivery of the intended dose prescription and also provides further opportunities to develop adaptive radiotherapy applications.

We are currently working to apply these algorithms for advanced clinical applications such as incorporating 4D CT datasets, non-coplanar VMAT delivery, and gated delivery methods.

We actively collaborate in this research area with Dr. Peter Greer of the University of Newcastle, Australia, to advance portal image dosimetry.  For example, some of our algorithms are incorporated into their ‘WatchDog’ project which seeks to identify gross radiotherapy delivery errors in real-time through the comparison of predicted portal transmission images with actual acquired portal transmission images. 

We are developing advanced x-ray scatter prediction techniques that combine Monte Carlo simulation with analytical methods to produce fast yet accurate estimates of scattered photon fluence entering imaging detectors.  Removal of this fluence component ahead of CT image reconstruction results in more accurate patient density maps which is beneficial for improved radiation targeting and adaptive radiation therapy techniques.

We are incorporating multi-parametric magnetic resonance imaging techniques to improve the management and treatment of prostate cancer.  These techniques include cutting-edge functional MRI applications.  Furthermore, we are working to establish MRI as a primary imaging modality for some radiotherapy simulation procedures, which would improve the system efficiency and the quality of delivered treatment.

The ability to track tumour motion in real-time and adjust the radiation treatment delivery to compensate for this motion is a significant challenge in radiotherapy.  We are developing techniques for markerless tracking of lung tumours, together with tumour motion prediction algorithms to allow the multi-leaf collimation available on radiation delivery machines to follow tumour motion in near real-time.  This approach will provide significant patient benefit in terms of improved dose delivered to the tumour combined with reduced treatment toxicity.

Optimization methods are widely used in radiation therapy.  We are investigating the use of multi-objective optimization methods, specifically evolutionary (or ‘genetic’) algorithms, for their usefulness in radiotherapy applications.  To date, radiotherapy has been dominated by classic, single-objective function optimization techniques, which require user-defined weighting parameters to define the importance of competing objectives.  In contrast, true multi-objective optimization techniques return multiple solutions that are mathematically equivalent and lay on the ‘Pareto front’, which is the surface of solutions where no one objective can be improved without degrading at least one other competing objective.  The user can then select the most appropriate solution from the set of Pareto solutions.  Multi-objective optimization is useful and necessary when dealing with non-convex problems, such as beam fluence optimization when the beam angle/trajectory is included in the problem, since single-objective optimization methods may not identify the global minima.  The medical physics group at CancerCare Manitoba collaborates on this topic with Professor Jason Fiege, an astrophysicist at the University of Manitoba, who utilizes evolutionary algorithms to tackle gravitational lensing and other large scale astrophysical problems.

We are investigating methods to utilize our knowledge of the delivered patient three-dimensional dosimetry (obtained via in vivo portal dosimetry methods) and adapt future treatment delivery to compensate for any identified deficiencies in the previously delivered treatments.  Adaptation techniques being examined include off-line, pre-treatment, and on-treatment options.  These methods will adapt the radiation therapy to account for on-treatment changes in anatomy or linear accelerator output, thus ensuring higher quality treatment delivery to the patient.

The Monte Carlo simulation method is the most accurate method available to solve radiation transport problems.  It may be used for the calculation of absorbed dose and for simulating conditions that are often experimentally difficult (if not impossible) to verify. This technique uses random numbers to simulate the transport of radiation within an object of interest. It has many applications including modeling the radiation beams produced by linear accelerators and diagnostic x-ray equipment, as well as radiation transport and dose deposition through phantoms, patients, and imaging systems.

The EGSnrc and BEAM Monte Carlo codes are used widely within the Medical Physics Department to study diverse topics such as scattered photon fluence prediction for Portal Imaging Dosimetry and ConeBeam Computed Tomography.

Building on the availability of detectors capable of detecting diagnostic energy photons with high spatial and energy resolving capabilities, we are developing new imaging algorithms which make use of scattered x-ray photons and microwaves to improve the sensitivity and specificity of malignant breast tumour detection.