Object Segmentation with Deep Neural Nets Coupled with a Shape Prior, When Learning From a Training set of Limited Quality and Small Size | IEEE Conference Publication | IEEE Xplore

Object Segmentation with Deep Neural Nets Coupled with a Shape Prior, When Learning From a Training set of Limited Quality and Small Size


Abstract:

Statistical shape priors can be crucial in segmenting objects when the data differentiates poorly between the object and its surroundings. For reliable learning, while so...Show More

Abstract:

Statistical shape priors can be crucial in segmenting objects when the data differentiates poorly between the object and its surroundings. For reliable learning, while some methods need high-quality expert segmentations, other methods need large training sets, both of which can often be difficult to obtain in clinical deployment or scientific studies. We propose to couple deep neural networks with a pointset-based shape prior that can be learned effectively despite training sets having small size and imperfections in expert curation. The prior relies on sparse Riemannian modeling in Kendall shape space. Results on clinical brain magnetic resonance imaging data show that our framework improves over the state of the art in segmenting the thalamus and the caudate.
Date of Conference: 03-07 April 2020
Date Added to IEEE Xplore: 22 May 2020
ISBN Information:

ISSN Information:

Conference Location: Iowa City, IA, USA

References

References is not available for this document.