Output details
11 - Computer Science and Informatics
University College London
Automatic Segmentation and Quantitative Analysis of the Articular Cartilages From Magnetic Resonance Images of the Knee
<23>Originality: introduces a fully automatic scheme to segment MR image of the human knee. The framework is the first to date to provide a fully automatic segmentation of the entire structure (3 cartilages, 3 bones).
Significance: a novel multi-component statistical shape model is presented, combined with deformable model and local Bayesian classification to provide extra refinement of the segmentation. The framework has been patented world-wide and been licensed to Siemens..
Rigour: the technique was tested against other existing techniques and proved to be the most accurate and robust. The testing was done using a large dataset of manually segmented knees.