Marginal Space Deep Learning: Efficient Architecture for Volumetric Image Parsing
- 7 March 2016
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Imaging
- Vol. 35 (5), 1217-1228
- https://doi.org/10.1109/tmi.2016.2538802
Abstract
Robust and fast solutions for anatomical object detection and segmentation support the entire clinical workflow from diagnosis, patient stratification, therapy planning, intervention and follow-up. Current state-of-the-art techniques for parsing volumetric medical image data are typically based on machine learning methods that exploit large annotated image databases. Two main challenges need to be addressed, these are the efficiency in scanning high-dimensional parametric spaces and the need for representative image features which require significant efforts of manual engineering. We propose a pipeline for object detection and segmentation in the context of volumetric image parsing, solving a two-step learning problem: anatomical pose estimation and boundary delineation. For this task we introduce Marginal Space Deep Learning (MSDL), a novel framework exploiting both the strengths of efficient object parametrization in hierarchical marginal spaces and the automated feature design of Deep Learning (DL) network architectures. In the 3D context, the application of deep learning systems is limited by the very high complexity of the parametrization. More specifically 9 parameters are necessary to describe a restricted affine transformation in 3D, resulting in a prohibitive amount of billions of scanning hypotheses. The mechanism of marginal space learning provides excellent run-time performance by learning classifiers in clustered, high-probability regions in spaces of gradually increasing dimensionality. To further increase computational efficiency and robustness, in our system we learn sparse adaptive data sampling patterns that automatically capture the structure of the input. Given the object localization, we propose a DL-based active shape model to estimate the non-rigid object boundary. Experimental results are presented on the aortic valve in ultrasound using an extensive dataset of 2891 volumes from 869 patients, showing significant improvements of up to 45.2% over the state-of-the-art. To our knowledge, this is the first successful demonstration of the DL potential to detection and segmentation in full 3D data with parametrized representations.Keywords
This publication has 33 references indexed in Scilit:
- Deep Feature Learning for Knee Cartilage Segmentation Using a Triplanar Convolutional Neural NetworkLecture Notes in Computer Science, 2013
- Complete valvular heart apparatus model from 4D cardiac CTMedical Image Analysis, 2012
- Reducing the Dimensionality of Data with Neural NetworksScience, 2006
- A Fast Learning Algorithm for Deep Belief NetsNeural Computation, 2006
- SPASM: A 3D-ASM for segmentation of sparse and arbitrarily oriented cardiac MRI dataMedical Image Analysis, 2006
- Learning to detect natural image boundaries using local brightness, color, and texture cuesIeee Transactions On Pattern Analysis and Machine Intelligence, 2004
- Active shape model segmentation with optimal featuresIEEE Transactions on Medical Imaging, 2002
- 3-D active appearance models: segmentation of cardiac MR and ultrasound imagesIEEE Transactions on Medical Imaging, 2002
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998
- Active Shape Models-Their Training and ApplicationComputer Vision and Image Understanding, 1995