DOS 513 - Week 1 Discussion
Initial Post: Protons and Photons and CyberKnife, Oh My!
As a proton facility, we absolutely MUST use inhomogeneity correction factors when planning anywhere in the body, and especially in the thoracic area. This is because of the property of proton beam interactions with matter that causes them to penetrate to a predictable depth and then stop.1 This penetration depth is calculated based on the stopping power of the tissue that the beam encounters along its path. Since we are trying to send a beam to a specific depth (the distal side of the target) and no further, we need an accurate calculation of the tissue's stopping power, which can only be derived if we take tissue inhomogeneity into account and correct for it. If we treated the entire body as having uniform density, we would calculate drastically wrong path lengths. Our current planning system, XiO, uses a pencil beam algorithm to calculate the proton dose distribution with tissue inhomogeneity factors incorporated into the calculation.
The thoracic cavity is somewhat of a challenge for us because lung tissue has relatively little stopping power compared to solid tissue, and so the proton beam will travel much further in lung than in solid tissue. Since we have to incorporate margins for motion, our target will almost always include low-density portions of the lung. The beam will have a tendency to keep going until it hits the far side of the lung, and stop shortly once it has re-entered solid tissue again. We have to be conscious of this when choosing beam angles, since overshoot may end up involving sensitive tissues. This overshoot will still usually be much less dose than if we were using x-rays, but it is not zero dose, which is the goal in most proton treatments.
Interestingly, we do actually plan for photon treatments for some of our patients because insurance companies often want documentation of why proton therapy is advantageous and worth the extra cost. In these cases we create both a proton plan and a photon plan and an accompanying report comparing the two. When we use XiO for x-ray beam planning, we have several inhomogeneity correction factor algorithms to choose from, and we usually select superposition because it provides good accuracy while still completing calculations in a reasonable amount of time.
In a conversation about algorithms in various planning systems with one of our physicists, Sara St. James (November 19, 2014), she made special note of CyberKnife's algorithm options for tissue inhomogeneity correction. These same concerns had previously been related to me by our Chief Physicist Tony Wong earlier in the year, but I had not had sufficient understanding of algorithms at that time to grasp their significance. There is an option to commission CyberKnife with the highly accurate Monte Carlo algorithm for tissue inhomogeneity correction, but because of the long calculation time, many clinics opt for the simpler and faster option, which is a basic ray-tracing algorithm with path length correction, but no accounting for lateral scattering differences due to nearby structures of differing density.2 This simpler algorithm is effective and fairly accurate for treatment sites where tissues are in fact relatively homogenous, such as the brain, but the lungs are full of inhomogeneities created by bronchioles and blood vessels, not to mention potential irregularities in the shape of the tumor. Sharma et al showed that lung plans calculated to have adequate target coverage with the ray tracing technique typically had 28% +/- 15% less target coverage than expected when recalculated with the more accurate Monte Carlo technique. That's a pretty strong argument for having a bit of patience to wait for the Monte Carlo result.
- Khan FM. The Physics of Radiation Therapy. Lippincott Williams & Wilkins; 2012.
- Sharma SC, Ott JT, Williams JB, Dickow D. Clinical implications of adopting Monte Carlo treatment planning for CyberKnife. J Appl Clin Med Phys. 2010;11(1):3142.