Plan assessment metrics for dose painting in stereotactic radiosurgery

Abstract

PURPOSE As radiotherapy treatment precision increases with advancements in imaging and radiation delivery, dose painting treatment becomes increasingly feasible, where targets receive a non-uniform radiation dose. The high precision of stereotactic radiosurgery (SRS) makes it a good candidate for dose painting treatments, but no suitable metrics to assess dose painting SRS plans exist. Existing dose painting assessment metrics weight target overdose and underdose equally, but are unsuited for SRS plans which typically avoid target underdose more. Current SRS metrics also prioritise reducing healthy tissue dose through selectivity and dose fall-off, and these metrics assume single prescriptions. We propose a set of metrics for dose painting SRS that would meet clinical needs and are calculated with non-uniform dose painting prescriptions. METHODS & MATERIALS Sample dose painting SRS prescriptions are first created from Gamma Knife SRS cases, apparent diffusion coefficient MRI images, and various image-to-prescription functions. Treatment plans are found through semi-infinite linear programming optimisation and using clinically determined isocentres, then assessed with existing and proposed metrics. Modified versions of SRS metrics are proposed, including coverage, selectivity, conformity, efficiency, and gradient indices. Quality factor, a current dose painting metric, is applied both without changes and with modifications. A new metric, integral dose ratio, is proposed as a measure of target overdose. RESULTS The merits of existing and modified metrics are demonstrated and discussed. A modified conformity index using mean or minimum prescription dose would be suitable for dose painting SRS with integral or maximum boost methods respectively. Either modified efficiency index is a suitable replacement for the existing gradient index. CONCLUSIONS The proposed modified SRS metrics are appropriate measures of plan quality for dose painting SRS plans, and have the advantage of giving equal values as the original SRS metrics when applied to single-prescription plans.