Discretely-constrained deep network for weakly supervised segmentation.

Journal: Neural networks : the official journal of the International Neural Network Society
Published Date:

Abstract

An efficient strategy for weakly-supervised segmentation is to impose constraints or regularization priors on target regions. Recent efforts have focused on incorporating such constraints in the training of convolutional neural networks (CNN), however this has so far been done within a continuous optimization framework. Yet, various segmentation constraints and regularization priors can be modeled and optimized more efficiently in a discrete formulation. This paper proposes a method, based on the alternating direction method of multipliers (ADMM) algorithm, to train a CNN with discrete constraints and regularization priors. This method is applied to the segmentation of medical images with weak annotations, where both size constraints and boundary length regularization are enforced. Experiments on two benchmark datasets for medical image segmentation show our method to provide significant improvements compared to existing approaches in terms of segmentation accuracy, constraint satisfaction and convergence speed.

Authors

  • Jizong Peng
    Department of Software and IT Engineering, ETS Montreal, 1100 Notre-Dame W., Montreal, H3C 1K3, Canada. Electronic address: jizong.peng.1@etsmtl.ca.
  • Hoel Kervadec
    ÉTS Montréal, QC, Canada. Electronic address: hoel.kervadec.1@etsmtl.net.
  • Jose Dolz
    AQUILAB, Biocentre A. Fleming, 250 rue Salvador Allende, 59120, Loos les Lille, France. jose.dolz.upv@gmail.com.
  • Ismail Ben Ayed
    LIVIA Laboratory, École de technologie supérieure (ETS), Montreal, QC, Canada.
  • Marco Pedersoli
    Department of Automated Production, ETS Montreal, 1100 Notre-Dame W., Montreal, H3C 1K3, Canada.
  • Christian Desrosiers
    LIVIA Laboratory, École de technologie supérieure (ETS), Montreal, QC, Canada.