P317. Volume measurements and 3D modelling of Crohn's perianal fistulas – a novel technique
N.A. Yassin1, P.F. Lung2, A. Gupta2, R.K.S. Phillips1, P.E. Edwards3, A.L. Hart4, 1St Mark's Hospital and Academic Institute, Harrow, The Department of Colorectal Surgery, London, United Kingdom, 2St Mark's Hospital and Academic Institute, Harrow, The Department of Radiology, London, United Kingdom, 3Department of Surgical Imaging, Imperial College London, United Kingdom, 4IBD Unit, St Mark's Hospital and Academic Institute, Harrow, London, United Kingdom
Fistulating perianal Crohn's disease (CD) is a challenging condition that requires combined medical and surgical management, guided by radiology. Magnetic Resonance Imaging (MRI) is the gold standard assessment tool which is used to assess response to medical therapy. The use of anti-TNF therapy for the treatment of these fistulas relies heavily on clinical and radiological assessments in order to assess response. Clinical healing has been shown to lag behind radiological healing of Crohn's perianal fistulae, and frequent MRI follow up is part of the management protocol for these patients as it aids in the decisions with regards to biological therapy. It is difficult to predict response based on MRI images as there are currently no objective parameters. Having a 3D fistula model and being able to quantify the changes in the fistula volume would be beneficial in guiding both medical and surgical management.
The aim of this study was to test the use of computer software and MRI segmentation in order to design 3D models of anal fistula tract(s) and measure fistula volumes from MRI images.
MRI images of high and low volume, simple or complex fistulas were randomly selected. 3D fistula models were created using validated surgical computer software. Manual and semi-automated fistula volumes were calculated by the surgical trainee and a blinded radiology trainee. Variability between the techniques, and inter-observer variability were assessed.
Manual and semi-automated volumes of ten 3D fistula models were compared using the Wilcoxon matched-pairs test. For the surgeon, a median fistula volume of 7143 mm3 (IQR 1638, 12020) was noted using the manual technique, and 7772 mm3 (IQR 1442, 15500) using the semi-automated technique. For the radiologist the median volume using the manual technique was 7552 mm3 (IQR 1813, 13218) and 7782 mm3 (IQR 2006, 13723) for the semi-automated technique. There were no significant differences between the two methods for the surgeon (P = 0.17) or radiologist (P = 0.45).
The intra-class correlation coefficient showed that all variation in the measurements were between the different patients (97% surgeon, 99% radiologist), with minimal variation between the two methods of volume measurements (3% and 1%).
MRI volume measurements of 3D models of perianal fistula tracts provide a reproducible assessment tool as an adjunct to clinical follow up. Serial volume measurements are feasible and will guide the use of medical therapies such as anti-TNFs.