Jonathan Taylor, Presenter: Nothing to Disclose

Peter Metherall PhD, Abstract Co-Author: Nothing to Disclose

David Barber, Abstract Co-Author: Nothing to Disclose

Automatic VOI warping through non-linear registration has been shown to correlate well with contouring techniques in most cases. Further work needs to be done to verify that the tumour growth identified corresponds to a real increase in tumour burden. The registration algorithm needs to be further refined to cope with less well defined tumour boundaries. This technique could be transferrable to other types of serial studies References: [1] Rees et al. 2009. Volumes and growth rates of untreated adult low-grade gliomas indicate risk of early malignant transformation, EJR, 72(1) [2] Barber,Hose 2005. Automatic segmentation of medical images using image registration. JME & T, 29(2)  

Accurate quantification of LGG volume growth rate is important in determining the risk of malignant transformation [1]. At Sheffield 3D lab LGG volume is determined from MRI data using semi-automatic contouring methods, repeated at least 3 times by at least 2 different operators for each scan. An investigation was carried out to determine whether non-linear registration could be used to deform contours in order to automatically measure changes in tumour volume over time. This could potentially reduce variability in results and reduce processing time

Baseline tumour volume was determined by taking an average of the Volumes of Interest (VOIs) generated through contouring in MIM software. Consecutive MRI-FLAIR datasets were then non-linearly registered together using the Sheffield Image Registration Toolkit (ShIRT [2]). The generated registration map was applied to the averaged VOI to create a second, warped VOI. The volume of the warped VOI was compared to that measured through semi-automatic contouring methods on the second image. 10 patients have been processed in this way so far

In cases where tumour volume growth was not overly large (<250% increase in size, n=9) the warped VOI was similar in shape and size to those generated from semi-automatic contouring methods. Volume from the automatic method was within 2S.D of that measured through semi-automatic methods in 7 cases

Taylor, J, Metherall, P, Barber, D, Use of Non-linear Registration to Track Changes in the Volume of Low Grade Gliomas (LGGs).  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12028316.html