Automatic Optimum Atlas Selection for Multi-Atlas Image Segmentation using Joint Label Fusion

. . . . . . . . . . . . . . . . . . . . . . . . . . 4 2 Background and Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.1 Manual image segmentation 2.2 Automatic image segmentation 2.3 Multi-atlas image segmentation 2.4 Label Fusion 2.5 Atlas selection 2.6 Automatic Optimum Atlas Selection (OAS) 3 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 3.1 Data and setup 3.1.1 Magnetic Resonance (MR) brain images dataset 3.1.2 Atlases and labels dataset 3.1.3 Computing and software infrastructure 3.2 OAS via global majority voting fusion image similarity score selection criteria 3.2.1 Atlas and target image pre-processing 3.2.2 Image registration and transformation 3.2.3 Global voting fusion image Optimum Atlas Selection (OAS) Framework 3.2.3.1 Consensus majority voting fusion segmentation 3.2.3.2 Dice similarity coefficient calculation (dSc) 3.2.3.3 Atlas selection criteria 3.2.4 OAS plus joint label fusion implementation 3.2.4.1 JLF Alpha (α) and Beta (β) parameters optimization 3.2.4.2 JLF search (rs) and patch (rp) size optimization 3.3 Leave-one out validation 4 Schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 5.1 Optimum atlas selection framework 5.1.1 Optimum image registration and label fusion image pad size 5.1.2 Image registration and transformation parameters optimization 5.1.3 Global Voting Optimum Atlas Selection (OAS) 5.1.3.1 Consensus Majority Segmentation 5.1.3.2 Atlas Selection 5.1.4 OAS plus joint label fusion implementation 5.1.4.1 Alpha (α) and Beta (β) parameters optimization 5.1.4.2 Search (rs) and patch (rp) radii size parameters optimization 5.1.4.3 Optimum Atlas Selection Result (Segmentation of LM Head MR Image) 5.2 Leave-one out validation 6 Discussion and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Automatic Optimum Atlas Selection for Multi-Atlas Image Segmentation using Joint Label Fusion Kofi Agyeman Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 7 Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Appendix A – Table of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Appendix B Software installation instructions . . . . . . . . . . . . . . . . . . 30 Appendix C – Scripts . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 Automatic Optimum Atlas Selection for Multi-Atlas Image Segmentation using Joint Label Fusion Kofi Agyeman