Analyzing patient specific data using gamma index

Pre-treatment verifications and on-line measurements are traditionally analyzed using gamma index. Below I address some of the drawbacks of using a gamma index and possible other solutions that can be both fast and clinically relevant.

Gamma index

The Gamma index was designed to be an index that is sensitive enough in the region where the dose doesn’t change too much and at the same time doesn’t get too sensitive in the gradient region. It thereby combines dose deviation with distance to agreement where distance to agreement is the distance from a specific measured dose to the nearest point with the same planned dose. It is easy to understand that in a gradient region e.g. penumbra a small distance changes the dose extensively and the vice versa in the flat region.

Gamma index is an absolute value and thereby it doesn’t tell if the dose is exceeded or too low. This doesn’t matter as long as we don’t know where the specific measurement point is located in the patient anatomy. However, when we do know where it is located it has a major impact, an over-dose in the tumor is not of the same clinical relevance as an under-dose. And similarly an under-dose in an OAR is nothing we would be to anxious to correct for compared to the treatment plan.

Dose discrepancy in patient anatomy

One solution to make the analysis more clinical relevant is to convert data into dose in patient anatomy and define if a structure is a target or OAR. The gamma value for each specific point can thereby be based on its location and take into account for positive and negative doses and only indicate discrepancies that are clinically relevant for each specific structure.

The drawback on this solution is that the calculation takes some time, and the answer is thereby not instant after measurement. It also requires to bring in the CT information into the measurement system and the 3D dose distribution also in the patient anatomy. To perform this additional steps for all patients is time consuming. The additional required time for this analysis reduces the use of this more clinical relevant procedure.

Practical clinical relevant solution

A practical solution to the above is offered by ScandiDos in the Delta4 DVH professional software option. In this software the patient structures are overlaid with the phantom and the measurement points are thereby grouped to what structure they belong. The same approach as in the patient anatomy case can thereby be applied distinguishing between over-dose and under-dose based on clinical relevance.

This solution has limitations because a variation in anatomy is not taken into account. Nevertheless,  when comparing with the treatment plan at the same conditions this limitation is rather small. This solution also has several distinct benefits. First, the data is received instantly and readily available. Second, there is no reduction in accuracy due to calculation in anatomy. The third benefit is that less data has to be exported from the TPS.

Due to the benefits and the small limitation the phantom-structure model is preferred by many users.

Görgen Nilsson,
Founder and CTO ScandiDos AB