-
[show abstract]
[hide abstract]
ABSTRACT: Landmark-based 3D hippocampal shape classification involves high-dimensional descriptor space, many noisy and redundant features,
and a very small number of training samples. Feature selection becomes critical in this situation, because it not only improves
classification performance, but also identifies the regions that contribute more to shape discrimination. This work identifies
the drawbacks of SVM-RFE, and proposes a novel class-separability-based feature selection approach to overcome them. We formulate
feature selection as a constrained integer optimization and develop a new algorithm to efficiently and optimally solve this
problem. Theoretical analysis and experimental study on both synthetic data and real hippocampus data demonstrate its superior
performance over the prevailing SVM-RFE. Our work provides a new efficient feature selection tool for hippocampal shape classification.
09/2010: pages 266-273;
-
[show abstract]
[hide abstract]
ABSTRACT: Scatter-matrix-based class separability is a simple and efficient feature selection criterion in the literature. However, the conventional trace-based formulation does not take feature redundancy into account and is prone to selecting a set of discriminative but mutually redundant features. In this brief, we first theoretically prove that in the context of this trace-based criterion the existence of sufficiently correlated features can always prevent selecting the optimal feature set. Then, on top of this criterion, we propose the redundancy-constrained feature selection (RCFS). To ensure the algorithm's efficiency and scalability, we study the characteristic of the constraints with which the resulted constrained 0-1 optimization can be efficiently and globally solved. By using the totally unimodular (TUM) concept in integer programming, a necessary condition for such constraints is derived. This condition reveals an interesting special case in which qualified redundancy constraints can be conveniently generated via a clustering of features. We study this special case and develop an efficient feature selection approach based on Dinkelbach's algorithm. Experiments on benchmark data sets demonstrate the superior performance of our approach to those without redundancy constraints.
IEEE Transactions on Neural Networks 06/2010; · 2.95 Impact Factor
-
Medical Image Computing and Computer-Assisted Intervention - MICCAI 2010, 13th International Conference, Beijing, China, September 20-24, 2010, Proceedings, Part II; 01/2010
-
IEEE Transactions on Neural Networks. 01/2010; 21:853-858.
-
[show abstract]
[hide abstract]
ABSTRACT: Landmark-based 3D hippocampal shape classification involves high-dimensional descriptor space, many noisy and redundant features, and a very small number of training samples. Feature selection becomes critical in this situation, because it not only improves classification performance, but also identifies the regions that contribute more to shape discrimination. This work identifies the drawbacks of SVM-RFE, and proposes a novel class-separability-based feature selection approach to overcome them. We formulate feature selection as a constrained integer optimization and develop a new algorithm to efficiently and optimally solve this problem. Theoretical analysis and experimental study on both synthetic data and real hippocampus data demonstrate its superior performance over the prevailing SVM-RFE. Our work provides a new efficient feature selection tool for hippocampal shape classification.
Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention. 01/2010; 13(Pt 2):266-73.
-
[show abstract]
[hide abstract]
ABSTRACT: In this article, we present a framework to perform statistical shape analysis for segmented hippocampi, including an efficient permutation test for detecting subtle class differences, and a regularized discriminative direction method for visualizing the shape discrepancy. Fisher permutation and bootstrap tests are preferred to traditional hypothesis tests which impose assumptions on the distribution of the samples. In this article, an efficient algorithm is adopted to rapidly perform the exact tests. We extend this algorithm to multivariate data by projecting the shape descriptors onto an informative direction that preserves the original discriminative information as much as possible to generate a scalar test statistic. This informative direction is further used to seek a discriminative direction to isolate the discriminative shape difference between classes from the individual variability. Compared with existing methods, the discriminative direction used in this article is regularized by requiring that the shapes deformed along it respect the underlying shape distribution as well as reflecting the essential shape differences between two populations. Hence, a more accurate localization of difference is produced. We apply our methods to analyze the hippocampal shapes for controls and subjects with Alzheimer's disease from the publicly available OASIS MRI database. We show how to localize the shape differences between the two classes.
Hippocampus 06/2009; 19(6):533-40. · 5.18 Impact Factor
-
IEEE Trans. Med. Imaging. 01/2009; 28:937-950.
-
[show abstract]
[hide abstract]
ABSTRACT: The “discriminative direction” has been proven useful to reveal the subtle difference between two anatomical shape classes.
When a shape moves along this direction, its deformation will best manifest the class difference detected by a kernel classifier.
However, we observe that such a direction cannot maintain a shape’s “anatomical” correctness, introducing spurious difference.
To overcome this drawback, we develop a regularized discriminative direction by requiring a shape to conform to its population distribution when it deforms along the discriminative
direction. Instead of iterative optimization, an analytic solution is provided to directly work out this direction. Experimental
study shows its superior performance in detecting and localizing the difference of hippocampal shapes for sex. The result
is supported by other independent research in the same domain.
10/2008: pages 628-635;
-
[show abstract]
[hide abstract]
ABSTRACT: The "discriminative direction" has been proven useful to reveal the subtle difference between two anatomical shape classes. When a shape moves along this direction, its deformation will best manifest the class difference detected by a kernel classifier. However, we observe that such a direction cannot maintain a shape's "anatomical" correctness, introducing spurious difference. To overcome this drawback, we develop a regularized discriminative direction by requiring a shape to conform to its population distribution when it deforms along the discriminative direction. Instead of iterative optimization, an analytic solution is provided to directly work out this direction. Experimental study shows its superior performance in detecting and localizing the difference of hippocampal shapes for sex. The result is supported by other independent research in the same domain.
Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention. 02/2008; 11(Pt 1):628-35.
-
Medical Image Computing and Computer-Assisted Intervention - MICCAI 2008, 11th International Conference, New York, NY, USA, September 6-10, 2008, Proceedings, Part I; 01/2008
-
Computer Vision - ECCV 2008, 10th European Conference on Computer Vision, Marseille, France, October 12-18, 2008, Proceedings, Part IV; 01/2008
-
[show abstract]
[hide abstract]
ABSTRACT: Hypothesis testing is an important way to detect the statistical difference between two populations. In this paper, we use the Fisher permutation and bootstrap tests to differentiate hippocampal shape between genders. These methods are preferred to traditional hypothesis tests which impose assumptions on the distribution of the samples. An efficient algorithm is adopted to rapidly perform the exact tests. We extend this algorithm to multivariate data by projecting the original data onto an "informative direction" to generate a scalar test statistic. This "informative direction" is found to preserve the original discriminative information. This direction is further used in this paper to isolate the discriminative shape difference between classes from the individual variability, achieving a visualization of shape discrepancy.
Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention. 02/2007; 10(Pt 1):375-83.
