Interpretable exemplar-based shape classification using constrained sparse linear models

Gunnar A. Sigurdsson, Zhen Yang, Trac D. Tran, Jerry Ladd Prince

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Many types of diseases manifest themselves as observable changes in the shape of the affected organs. Using shape classification, we can look for signs of disease and discover relationships between diseases. We formulate the problem of shape classification in a holistic framework that utilizes a lossless scalar field representation and a non-parametric classification based on sparse recovery. This framework generalizes over certain classes of unseen shapes while using the full information of the shape, bypassing feature extraction. The output of the method is the class whose combination of exemplars most closely approximates the shape, and furthermore, the algorithm returns the most similar exemplars along with their similarity to the shape, which makes the result simple to interpret. Our results show that the method offers accurate classification between three cerebellar diseases and controls in a database of cerebellar ataxia patients. For reproducible comparison, promising results are presented on publicly available 2D datasets, including the ETH-80 dataset where the method achieves 88.4% classification accuracy.

Original languageEnglish (US)
Title of host publicationMedical Imaging 2015: Image Processing
PublisherSPIE
Volume9413
ISBN (Print)9781628415032
DOIs
StatePublished - 2015
EventMedical Imaging 2015: Image Processing - Orlando, United States
Duration: Feb 24 2015Feb 26 2015

Other

OtherMedical Imaging 2015: Image Processing
CountryUnited States
CityOrlando
Period2/24/152/26/15

Fingerprint

Linear Models
Cerebellar Diseases
Cerebellar Ataxia
ataxia
Feature extraction
Databases
pattern recognition
organs
Recovery
recovery
scalars
output
Datasets

Keywords

  • interpretable classifiers
  • morphology
  • shape classification
  • signed distance functions
  • sparse recovery

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics
  • Electronic, Optical and Magnetic Materials
  • Biomaterials
  • Radiology Nuclear Medicine and imaging

Cite this

Sigurdsson, G. A., Yang, Z., Tran, T. D., & Prince, J. L. (2015). Interpretable exemplar-based shape classification using constrained sparse linear models. In Medical Imaging 2015: Image Processing (Vol. 9413). [94130R] SPIE. https://doi.org/10.1117/12.2082141

Interpretable exemplar-based shape classification using constrained sparse linear models. / Sigurdsson, Gunnar A.; Yang, Zhen; Tran, Trac D.; Prince, Jerry Ladd.

Medical Imaging 2015: Image Processing. Vol. 9413 SPIE, 2015. 94130R.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Sigurdsson, GA, Yang, Z, Tran, TD & Prince, JL 2015, Interpretable exemplar-based shape classification using constrained sparse linear models. in Medical Imaging 2015: Image Processing. vol. 9413, 94130R, SPIE, Medical Imaging 2015: Image Processing, Orlando, United States, 2/24/15. https://doi.org/10.1117/12.2082141
Sigurdsson GA, Yang Z, Tran TD, Prince JL. Interpretable exemplar-based shape classification using constrained sparse linear models. In Medical Imaging 2015: Image Processing. Vol. 9413. SPIE. 2015. 94130R https://doi.org/10.1117/12.2082141
Sigurdsson, Gunnar A. ; Yang, Zhen ; Tran, Trac D. ; Prince, Jerry Ladd. / Interpretable exemplar-based shape classification using constrained sparse linear models. Medical Imaging 2015: Image Processing. Vol. 9413 SPIE, 2015.
@inproceedings{967febca3bb1482a8a129d706121c5f8,
title = "Interpretable exemplar-based shape classification using constrained sparse linear models",
abstract = "Many types of diseases manifest themselves as observable changes in the shape of the affected organs. Using shape classification, we can look for signs of disease and discover relationships between diseases. We formulate the problem of shape classification in a holistic framework that utilizes a lossless scalar field representation and a non-parametric classification based on sparse recovery. This framework generalizes over certain classes of unseen shapes while using the full information of the shape, bypassing feature extraction. The output of the method is the class whose combination of exemplars most closely approximates the shape, and furthermore, the algorithm returns the most similar exemplars along with their similarity to the shape, which makes the result simple to interpret. Our results show that the method offers accurate classification between three cerebellar diseases and controls in a database of cerebellar ataxia patients. For reproducible comparison, promising results are presented on publicly available 2D datasets, including the ETH-80 dataset where the method achieves 88.4{\%} classification accuracy.",
keywords = "interpretable classifiers, morphology, shape classification, signed distance functions, sparse recovery",
author = "Sigurdsson, {Gunnar A.} and Zhen Yang and Tran, {Trac D.} and Prince, {Jerry Ladd}",
year = "2015",
doi = "10.1117/12.2082141",
language = "English (US)",
isbn = "9781628415032",
volume = "9413",
booktitle = "Medical Imaging 2015: Image Processing",
publisher = "SPIE",

}

TY - GEN

T1 - Interpretable exemplar-based shape classification using constrained sparse linear models

AU - Sigurdsson, Gunnar A.

AU - Yang, Zhen

AU - Tran, Trac D.

AU - Prince, Jerry Ladd

PY - 2015

Y1 - 2015

N2 - Many types of diseases manifest themselves as observable changes in the shape of the affected organs. Using shape classification, we can look for signs of disease and discover relationships between diseases. We formulate the problem of shape classification in a holistic framework that utilizes a lossless scalar field representation and a non-parametric classification based on sparse recovery. This framework generalizes over certain classes of unseen shapes while using the full information of the shape, bypassing feature extraction. The output of the method is the class whose combination of exemplars most closely approximates the shape, and furthermore, the algorithm returns the most similar exemplars along with their similarity to the shape, which makes the result simple to interpret. Our results show that the method offers accurate classification between three cerebellar diseases and controls in a database of cerebellar ataxia patients. For reproducible comparison, promising results are presented on publicly available 2D datasets, including the ETH-80 dataset where the method achieves 88.4% classification accuracy.

AB - Many types of diseases manifest themselves as observable changes in the shape of the affected organs. Using shape classification, we can look for signs of disease and discover relationships between diseases. We formulate the problem of shape classification in a holistic framework that utilizes a lossless scalar field representation and a non-parametric classification based on sparse recovery. This framework generalizes over certain classes of unseen shapes while using the full information of the shape, bypassing feature extraction. The output of the method is the class whose combination of exemplars most closely approximates the shape, and furthermore, the algorithm returns the most similar exemplars along with their similarity to the shape, which makes the result simple to interpret. Our results show that the method offers accurate classification between three cerebellar diseases and controls in a database of cerebellar ataxia patients. For reproducible comparison, promising results are presented on publicly available 2D datasets, including the ETH-80 dataset where the method achieves 88.4% classification accuracy.

KW - interpretable classifiers

KW - morphology

KW - shape classification

KW - signed distance functions

KW - sparse recovery

UR - http://www.scopus.com/inward/record.url?scp=84943405471&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84943405471&partnerID=8YFLogxK

U2 - 10.1117/12.2082141

DO - 10.1117/12.2082141

M3 - Conference contribution

AN - SCOPUS:84943405471

SN - 9781628415032

VL - 9413

BT - Medical Imaging 2015: Image Processing

PB - SPIE

ER -