Segmentation of Filamentary Structured Objects

Learning to Boost Filamentary Segmentation



Lin Gu, Li Cheng @ Machine Learning For Bioimage Analysis Group BII, A*STAR,Singapore

  • Results

  • Base methods:
  • [1] Kernel Boost that learns discriminative convolutional filters based on the Gradient Boosting framework.
  • [2] Optimally Oriented Flux (OOF), an unsupervised method with manually designed filter that is widely used for delineating tubular structures.
  • [3] IUWT , an unsupervised method based on isotropic undecimated wavelet transform.
  • [4] Eigen, a multiscale Hessian-based method.
  • [5] T2T, a supervised method based on pixel classification, medial sub-tree generation, and global tree linking components.
  • Performance on the retinal testbeds DRIVE & STARE with three different base methods

Kernel Boost OOF IUWT
Base 76.29 & 78.77 69.97 & 72.86 64.13 & 72.20
Ours-nCnM 79.12 & 79.63 71.22 & 73.28 71.79 & 74.15
Ours-MnC 79.44 & 81.14 71.93 & 75.91 72.25 & 74.29
Ours 80.11 & 81.93 72.12 & 76.84 72.65 & 74.58
Here the modified F1 measure (%) is used. Ours-nCnM refers to the simplified variant of our approach without using completion fields and matting, while Ours-MnC instead refers to without completion field but with matting. Ours is the full version of our approach.
  • Performance on 2D neronal testbed NeuB1 with four different base methods

Kernel Boost OOF Eigen T2T
Base 84.74 63.50 63.94 66.49
Ours-nMnC 86.20 65.99 66.81 66.18
Ours-MnC 86.39 70.25 70.38 75.70
Ours 86.80 71.03 71.89 76.01
  • References :
  • [1] C. Becker, R. Rigamonti, V. Lepetit, and P. Fua. Supervised feature learning for curvilinear structure segmentation. In MICCAI, 2013.
    [2] M. Law and A. Chung. Three dimensional curvilinear structure detection using optimally oriented flux. In ECCV, 2008.
    [3] P. Bankhead, C. Scholfield, J. McGeown, and T. Curtis. Fast retinal vessel detection and measurement using wavelets and edge location refinement. PLoS ONE, 2012.
    [4] A. Frangi, W. Niessen, K. Vincken, and M. Viergever. Multiscale vessel enhancement filtering. In MICCAI, 1998.
    [5] S. Basu, A. Aksel, B. Condron, and S. Acton. Tree2Tree: Neuron segmentation for generation of neuronal morphology. In ISBI, 2010.