RCR10: A Method of Unwrapping 3D Complex Hollow Organs for Spatial Dose Surface Analysis

Alon Witztum1,Samantha Warren1,Mike Partridge1,Maria Hawkins1

1CRUK/MRC Oxford Institute for Radiation Oncology, University of Oxford, Oxford, UK

Presenting date: Tuesday 3 November
Presenting time: 13.10-14.00


To describe and evaluate a method of mapping 3D surface dose for hollow organs with complex geometry (duodenum) to a 2D plane. Methods have previously been presented for unwrapping tubular organs (oesophagus and rectum) for spatial modelling of toxicity. However, the complex geometry of the duodenum demands a new method for producing 2D dose surface maps (DSMs).


The organ is defined by Delaunay triangulation and voxelised into a 0.085 × 0.085 × 0.085 cm3 binary image. This is thinned using a distance transform resulting in a cluster of points approximately down the organ centre. A subset of points is manually selected and a central axis spline (CAS) fit. The CAS is discretised into 30 points and a ray cast from each point (angular step 12°, axial step 0.05 cm) on the plane normal to the vector to the next CAS point and the dose at the closest perimeter point recorded. These perimeter slices are "unwrapped" from the edge distal to the pancreas so the high dose region (proximal to tumour) falls in the centre of the dose map.


A complete methodology has been described that produces 2D DSMs for organs with simple or complex geometries. This method was successfully tested by extracting DSMs for 15 duodena, with one oesophagus illustrating application to simple geometries. Visual comparisons show that a 30 נ30 map provides sufficient resolution to view features of interest.


A new method for unwrapping surface dose in complex-shaped hollow organs has been developed and tested on 15 duodenal cases. The method is robust, requires minimal human interaction, and has been shown to be generalisable to simpler geometries (oesophagus). The 2D dose maps produced provide spatial dose distribution information which will be explored to create models that may improve toxicity prediction in treatments for locally-advanced pancreatic cancer.