Automatic Full Femur Segmentation from Computed Tomography Datasets Using an Atlas-Based Approach
Automatic segmentation of femurs in clinical computed tomography remains a challenge. Joints degraded by old age are a particularly challenging dataset to segment. The objective of this study is to evaluate existing methods and propose an alternative method for segmentation of femurs in clinical computed tomography datasets for joints degraded by old age. Bilateral hip computed tomography scans of three cadaveric specimens (six femurs) were available for this study. Deformable registration using an affine selection criterion was used for atlas-based segmentation. For comparison, the six femurs were also segmented with two graph-cut algorithms. An automatic graph-cut segmentation algorithm was only able to separate the femur from the pelvis in two of the six femurs due to a limitation of graph-cuts. The atlas-based method produced consistent automatic segmentations for all degraded joints. In conclusion, atlas-based femur segmentation performs considerably better than an automatic graph-cut algorithm when applied to degraded joints.
KeywordsAtlas-based segmentation Graph-cut Femur Computed tomography
The authors would like to thank NSERC CGS-M for funding, the Body Donation Program at the Gross Anatomy Laboratory for access to cadaveric specimens, and the individuals who graciously contributed their bodies. The authors would also like to thank Dr. Sonny Chan from the Department of Computer Science at the University of Calgary for guidance.
- 6.Pauchard, Y., Fitze, T., Browarnik, D., Eskandari, A., Pauchard, I., Enns-Bray, W., Pálsson, H., Sigurdsson, S., Ferguson, S., Harris, T., Gudnason, V., Helgason, B.: Interactive graph-cut segmentation for fast creation of finite element models from clinical CT data for hip fracture prediction. Comput. Methods Biomech. Biomed. Eng. 19(16), 1693–1703 (2016)CrossRefGoogle Scholar
- 7.Krčah, M., Székely, G., Blanc, R.: Fully automatic and fast segmentation of the femur bone from 3D-CT images with no shape prior. In: Proceedings of the 8th IEEE International Symposium on Biomedical Imaging - ISBI 2011, pp. 2087–2090. IEEE (2011)Google Scholar
- 11.Whitmarsh, T., Treece, G.M., Poole, K.E.S.: Automatic segmentation and discrimination of connected joint bones from CT by multi-atlas registration. In: Yao, J., Klinder, T., Li, S. (eds.) Computational Methods and Clinical Applications for Spine Imaging. LNCVB, vol. 17, pp. 199–207. Springer, Cham (2014). https://doi.org/10.1007/978-3-319-07269-2_17CrossRefGoogle Scholar