Evaluation and comparison of 3D intervertebral disc localization and segmentation methods for 3D T2 MR data: A grand challenge
Medical Image Analysis
Elsevier for MICCAI Society
Reason for embargo
The evaluation of changes in Intervertebral Discs (IVDs) with 3D Magnetic Resonance (MR) Imaging (MRI) can be of interest for many clinical applications. This paper presents the evaluation of both IVD localization and IVD segmentation methods submitted to the Automatic 3D MRI IVD Localization and Segmentation challenge, held at the 2015 International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI2015) with an on-site competition. With the construction of a manually annotated reference data set composed of 25 3D T2-weighted MR images acquired from two different studies and the establishment of a standard validation framework, quantitative evaluation was performed to compare the results of methods submitted to the challenge. Experimental results show that overall the best localization method achieves a mean localization distance of 0.8 mm and the best segmentation method achieves a mean Dice of 91.8%, a mean average absolute distance of 1.1 mm and a mean Hausdorff distance of 4.3 mm, respectively. The strengths and drawbacks of each method are discussed, which provides insights into the performance of different IVD localization and segmentation methods.
The paper is partially supported by the Swiss National Science Foundation Project No. 205321−157207/1. The acquisition of original images was supported by the Grant 14431/02/NL/SH 2 from the European Space Agency, grant 50WB0720 from the German Aerospace Center (DLR) and the CharitView the MathML source University Medical School Berlin. The work of team UNILJU was partially supported by the Slovenian Research Agency, under grants P2−0232, J2−5473, J7−6781 and J7−7118. The work of team ICL was partially funded by the Dunhill Medical Trust, R401/0215. I. López Andrade is supported by the Fundación Barrié. D. Forsberg was funded by the Swedish Innovation Agency (VINNOVA, grant 2014-01422). H. Chen and D. Qi were funded by Hong Kong RGC Fund (Project No. CUHK 412513). M. Urschler was partially funded by province of Styria, ABT08-22-T-7/2013-13, and D. Stern by Austrian Science Fund (FWF), P28078-N33. The work of team UNIQUE was partially supported under Australian Research Council’s linkage project funding scheme LP100200422. The competence center VRVis with the grant number 843272 is funded within the scope of COMET.
This is the author accepted manuscript. The final version is available from Elsevier via the DOI in this record.
Vol. 35, pp. 327 - 344
Place of publication