Multi-Channel Volumetric Neural Network for Knee Cartilage Segmentation in Cone-Beam CT
- 52 Downloads
Analyzing knee cartilage thickness and strain under load can help to further the understanding of the effects of diseases like Osteoarthritis. A precise segmentation of the cartilage is a necessary prerequisite for this analysis. This segmentation task has mainly been addressed in Magnetic Resonance Imaging, and was rarely investigated on contrast-enhanced Computed Tomography, where contrast agent visualizes the border between femoral and tibial cartilage. To overcome the main drawback of manual segmentation, namely its high time investment, we propose to use a 3D Convolutional Neural Network for this task. The presented architecture consists of a V-Net with SeLu activation, and a Tversky loss function. Due to the high imbalance between very few cartilage pixels and many background pixels, a high false positive rate is to be expected. To reduce this rate, the two largest segmented point clouds are extracted using a connected component analysis, since they most likely represent the medial and lateral tibial cartilage surfaces. The resulting segmentations are compared to manual segmentations, and achieve on average a recall of 0.69, which confirms the feasibility of this approach.
Unable to display preview. Download preview PDF.
- 1.Powers CM, Ward SR, Fredericson M, et al. Patellofemoral kinematics during weight-bearing and non-weight-bearing knee extension in persons with lateral subluxation of the patella: a preliminary study. J Orthop Sports Phys Ther. 2003;33(11):677–685.Google Scholar
- 2.Raj A, Vishwanathan S, Ajani B, et al. Automatic knee cartilage segmentation using fully volumetric convolutional neural networks for evaluation of osteoarthritis. In: ISBI 2018. IEEE; 2018. p. 851–854.Google Scholar
- 3.Tabrizi PR, Zoroofi RA, Yokota F, et al. Shape-based acetabular cartilage segmentation: application to CT and MRI datasets. Int J Comput Assist Radiol Surg. 2016;11(7):1247–1265.Google Scholar
- 4.Baniasadipour A, Zoroofi RA, Sato Y, et al. A fully automated method for segmentation and thickness map estimation of femoral and acetabular cartilages in 3D CT images of the hip. In: ISPA 2007; 2007. p. 92–97.Google Scholar
- 5.Myller KAH, Honkanen JTJ, Jurvelin JS, et al. Method for segmentation of knee articular cartilages based on contrast-enhanced CT images. Ann Biomed Eng. 2018;46(11):1756–1767.Google Scholar
- 6.Milletari F, Navab N, Ahmadi S. V-Net: fully convolutional neural networks for volumetric medical image segmentation. In: 3DV 2016; 2016. p. 565–571.Google Scholar
- 7.Duchi J, Hazan E, Singer Y. Adaptive subgradient methods for online learning and stochastic optimization. J Mach Learn Res. 2011;12(Jul):2121–2159.Google Scholar
- 8.Salehi SSM, Erdogmus D, Gholipour A. Tversky loss function for image segmentation using 3D fully convolutional deep networks. In: MLMI 2017. Springer; 2017. p. 379–387.Google Scholar
- 9.Maier AK, Syben C, Stimpel B, et al. Learning with known operators reduces maximum error bounds. Nat Mach Intell. 2019;1(8):373–380.Google Scholar
- 10.Gomez AN, Ren M, Urtasun R, et al. The reversible residual network: backpropagation without storing activations. In: Adv Neural Inf Process Syst; 2017. p. 2214–2224.Google Scholar