Dose surface maps of the heart identify dose sensitive regions
McWilliam, Alan Graham, L Dootson, C Abravan, Azadeh Van Herk, Marcel
McWilliam, Alan
Graham, L
Dootson, C
Abravan, Azadeh
Van Herk, Marcel
Citations
Altmetric:
Abstract
Purpose or Objective
It has become accepted that for lung cancer patients the
radiation dose to the heart is an independent predictor of
survival. Analysis of dosimetric effects on the heart have
investigated volumetric dose statistics, however potential
critical cardiac structures are present on the heart
surface. Volumetric parameters may not be optimal due to
the required inter-patient registration accuracy. In this
work we present a methodology for creating cardiac
surface dose maps and use these to identify cardiac
surface regions where excess dose results in worse patient
survival.
Material and Methods
648 cardiac surface maps were successfully created with a
polar coordinate system with the centre positioned at the
centre of mass for the heart contour of each slice in
turn. This accounted for the asymmetric nature of the
heart and can be considered a modified cylindrical
coordinate system. The radiotherapy dose for each
patient was sampled on this surface with the described
coordinate system. All hearts were normalised to the
same superior-inferior dimensions for analysis.
For validation of the mapping, and localisation of
dosimetric effects, the cardiac chambers were manually
delineated and mapped onto the surface maps for 20
patients. A point spread function (PSF) was fitted to the
blurred edge of the mapped chambers to quantify
uncertainty in the mapping process. To account for this
uncertainty, dose maps were blurred by a 2D-Guassian
function with width described by the PSF.
Permutation testing was used to identify regions where
excess dose resulted in worse patient survival with the Tmap
calculated. A threshold was set at the 99th percentile
of the T-map and the dose from the cardiac surface each
patient received extracted for analysis in a multi-variable
cox-proportional hazards survival model.
Results
All surface maps were blurred with 2D-Guassian filter of
size σφ=4.3° and σy = 1.3 units. Permutation testing for
patients dead and alive at 6,12,18 and 24 months showed
significant differences, p<0.001. The T-map for 18 months is included as figure 1 highlighting the highest significance
region at the base of the heart and extending into the right
and left atria. The mean dose to the region defined by the
99th percentile across all patients was 21.6Gy compared to
the mean dose to the heart of 12.7Gy. Table 1 shows the
multi-variable analysis where the dose to this region on the
heart surface is significantly associated with survival,
hazard ratio 1.014 per Grey, p=0.03, controlling for
covariates including tumour volume.
Conclusion We successfully created a modified cylindrical polar
coordinate system for mapping radiotherapy dose to the
heart. Surface maps identified a region near the base of
the heart where excess dose results in worse patient
survival. This region extends over the left and right atria
close to the path of the coronary arteries suggesting these
sub-structures are driving this effect.
Description
Date
2020
Publisher
Collections
Keywords
Type
Meetings and Proceedings
Citation
McWilliam A, Graham L, Dootson C, Abravan A, Van Herk M. OC-0197: Dose surface maps of the heart identify dose sensitive regions. Radiotherapy and Oncology . 2020 Nov;152:S98.