Publication

Structured learning of assignment models for neuron reconstruction to minimize topological errors

Conference Article

Conference

IEEE International Symposium on Biomedical Imaging (ISBI)

Edition

13th

Pages

607-611

Doc link

http://dx.doi.org/10.1109/ISBI.2016.7493341

File

Download the digital copy of the doc pdf document

Authors

Projects associated

Abstract

Structured learning provides a powerful framework for empirical risk minimization on the predictions of structured models. It allows end-to-end learning of model parameters to minimize an application specific loss function. This framework is particularly well suited for discrete optimization models that are used for neuron reconstruction from anisotropic electron microscopy (EM) volumes. However, current methods are still learning unary potentials by training a classifier that is agnostic about the model it is used in. We believe the reason for that lies in the difficulties of (1) finding a representative training sample, and (2) designing an application specific loss function that captures the quality of a proposed solution. In this paper, we show how to find a representative training sample from human generated ground truth, and propose a loss function that is suitable to minimize topological errors in the reconstruction. We compare different training methods on two challenging EM-datasets. Our structured learning approach shows consistently higher reconstruction accuracy than other current learning methods.

Categories

computer vision, object detection.

Author keywords

connectomics, structured learning, electron microscopy

Scientific reference

J. Funke, J. Klein, F. Moreno-Noguer, A. Cardona and M. Cook. Structured learning of assignment models for neuron reconstruction to minimize topological errors, 13th IEEE International Symposium on Biomedical Imaging, 2016, Prague, pp. 607-611.