Validation of deformable registration in adaptive radiation therapy with scale invariant feature transform

A critical issue in Deformable Registration (DR) is its validation due to the lack of a gold standard in clinical practice. We analyze Scale Invariant Feature Transform (SIFT) to quantify its potential as a tool for DR assessment. For this purpose we preliminary verify SIFT properties on both rigid and non-rigid phantoms and then apply it to evaluate DR in Head&Neck Adaptive Radiation Therapy. Results demonstrate SIFT capabilities as a performance assessment metric which is operator-independent and not limited to the analysis of specific structures.