Margins to account for cardiac and respiratory motion in cardiac radioablation

Abstract

Background

Cardiac radioablation (CR) is an emerging treatment for ventricular tachycardia, a rapid abnormal heart rhythm. Effectively delivering radiation to CR targets requires understanding and accounting for geometric uncertainties. One important uncertainty is motion induced by the cardiac and respiratory cycles, which can be accounted for by expanding the targeted region by a margin accounting for the motion's effect on dosimetry.

Purpose

To investigate margins to account for cardiac and respiratory motions in CR and compare different methods of computing these margins.

Methods

Eighty four hundred cardiorespiratory motion traces were created by joining 1050 cardiac motions derived from 30 Hz magnetic resonance images with eight respiratory motions from 5 Hz bi-planar kV fluoroscopy. Cardiac motions for each of the 17 segments of the left ventricle were acquired for 50 heart failure patients with a reduced ejection fraction. Respiratory motions were derived from the implantable cardioverter defibrillator lead's tip for eight CR patients. The margins needed to account for random errors were found using the convolution method by blurring a dose penumbra (Gaussian fall-off ${\sigma }_p$= 3.2 mm) with the motion. The motion margin was computed as the shift in the 95% dose level after blurring. Since these dosimetric margins do not consider rotations and shape deformation, they are considered a lower limit to account for cardiorespiratory motions. These motion margins were compared to (i) a sum of cardiac and respiratory motion amplitudes, similar to using an internal target volume (ITV); (ii) the van Herk et al. margin formula (MF = $\beta \left(\sigma -{\sigma }_p\right)$); and (iii) the amplitude of respiratory motion alone, similar to using a respiratory ITV.

Results

The sum of cardiac and respiratory motion amplitudes significantly overestimated the motion margins by [2.2±0.7 right-left, 2.6±0.9 ant-post, 2.7±0.7 inf-sup] mm. The margin formula accurately calculated the motion margins with average differences from the convolution method of [0.00±0.06, 0.0±0.1, 0.0±0.1] mm. Accounting for the amplitude of respiratory motion alone was on average sufficient but not robust, as it could underestimate the motion margin by up to 5 mm.

Conclusions

Margins to account for cardiorespiratory motion in CR can be calculated using a margin formula. The conservative approach of accounting for the amplitude of cardiorespiratory motion can significantly overestimate the needed margin which may result in excess healthy tissue damage.