Learning-based stochastic object models for use in optimizing imaging systems

Steven R. Dolly, Mark A. Anastasio, Lifeng Yu, Hua Li

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

Abstract

It is widely known that the optimization of imaging systems based on objective, or task-based, measures of image quality via computer-simulation requires use of a stochastic object model (SOM). However, the development of computationally tractable SOMs that can accurately model the statistical variations in anatomy within a specified ensemble of patients remains a challenging task. Because they are established by use of image data corresponding a single patient, previously reported numerical anatomical models lack of the ability to accurately model inter- patient variations in anatomy. In certain applications, however, databases of high-quality volumetric images are available that can facilitate this task. In this work, a novel and tractable methodology for learning a SOM from a set of volumetric training images is developed. The proposed method is based upon geometric attribute distribution (GAD) models, which characterize the inter-structural centroid variations and the intra-structural shape variations of each individual anatomical structure. The GAD models are scalable and deformable, and constrained by their respective principal attribute variations learned from training data. By use of the GAD models, random organ shapes and positions can be generated and integrated to form an anatomical phantom. The randomness in organ shape and position will reflect the variability of anatomy present in the training data. To demonstrate the methodology, a SOM corresponding to the pelvis of an adult male was computed and a corresponding ensemble of phantoms was created. Additionally, computer-simulated X-ray projection images corresponding to the phantoms were computed, from which tomographic images were reconstructed.

Original languageEnglish (US)
Title of host publicationMedical Imaging 2017
Subtitle of host publicationPhysics of Medical Imaging
PublisherSPIE
Volume10132
ISBN (Electronic)9781510607095
DOIs
StatePublished - 2017
EventMedical Imaging 2017: Physics of Medical Imaging - Orlando, United States
Duration: Feb 13 2017Feb 16 2017

Other

OtherMedical Imaging 2017: Physics of Medical Imaging
CountryUnited States
CityOrlando
Period2/13/172/16/17

Fingerprint

Imaging systems
learning
Anatomy
Learning
Anatomic Models
Aptitude
anatomy
Statistical Models
Pelvis
Computer Simulation
education
X-Rays
Databases
organs
methodology
pelvis
Image quality
Numerical models
centroids
computerized simulation

Keywords

  • Image quality assessment
  • Imaging system optimization
  • Stochastic object models

ASJC Scopus subject areas

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

Cite this

Dolly, S. R., Anastasio, M. A., Yu, L., & Li, H. (2017). Learning-based stochastic object models for use in optimizing imaging systems. In Medical Imaging 2017: Physics of Medical Imaging (Vol. 10132). [101321T] SPIE. https://doi.org/10.1117/12.2254055

Learning-based stochastic object models for use in optimizing imaging systems. / Dolly, Steven R.; Anastasio, Mark A.; Yu, Lifeng; Li, Hua.

Medical Imaging 2017: Physics of Medical Imaging. Vol. 10132 SPIE, 2017. 101321T.

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

Dolly, SR, Anastasio, MA, Yu, L & Li, H 2017, Learning-based stochastic object models for use in optimizing imaging systems. in Medical Imaging 2017: Physics of Medical Imaging. vol. 10132, 101321T, SPIE, Medical Imaging 2017: Physics of Medical Imaging, Orlando, United States, 2/13/17. https://doi.org/10.1117/12.2254055
Dolly SR, Anastasio MA, Yu L, Li H. Learning-based stochastic object models for use in optimizing imaging systems. In Medical Imaging 2017: Physics of Medical Imaging. Vol. 10132. SPIE. 2017. 101321T https://doi.org/10.1117/12.2254055
Dolly, Steven R. ; Anastasio, Mark A. ; Yu, Lifeng ; Li, Hua. / Learning-based stochastic object models for use in optimizing imaging systems. Medical Imaging 2017: Physics of Medical Imaging. Vol. 10132 SPIE, 2017.
@inproceedings{cb6b9c57b55e48af9360936e05460d7d,
title = "Learning-based stochastic object models for use in optimizing imaging systems",
abstract = "It is widely known that the optimization of imaging systems based on objective, or task-based, measures of image quality via computer-simulation requires use of a stochastic object model (SOM). However, the development of computationally tractable SOMs that can accurately model the statistical variations in anatomy within a specified ensemble of patients remains a challenging task. Because they are established by use of image data corresponding a single patient, previously reported numerical anatomical models lack of the ability to accurately model inter- patient variations in anatomy. In certain applications, however, databases of high-quality volumetric images are available that can facilitate this task. In this work, a novel and tractable methodology for learning a SOM from a set of volumetric training images is developed. The proposed method is based upon geometric attribute distribution (GAD) models, which characterize the inter-structural centroid variations and the intra-structural shape variations of each individual anatomical structure. The GAD models are scalable and deformable, and constrained by their respective principal attribute variations learned from training data. By use of the GAD models, random organ shapes and positions can be generated and integrated to form an anatomical phantom. The randomness in organ shape and position will reflect the variability of anatomy present in the training data. To demonstrate the methodology, a SOM corresponding to the pelvis of an adult male was computed and a corresponding ensemble of phantoms was created. Additionally, computer-simulated X-ray projection images corresponding to the phantoms were computed, from which tomographic images were reconstructed.",
keywords = "Image quality assessment, Imaging system optimization, Stochastic object models",
author = "Dolly, {Steven R.} and Anastasio, {Mark A.} and Lifeng Yu and Hua Li",
year = "2017",
doi = "10.1117/12.2254055",
language = "English (US)",
volume = "10132",
booktitle = "Medical Imaging 2017",
publisher = "SPIE",

}

TY - GEN

T1 - Learning-based stochastic object models for use in optimizing imaging systems

AU - Dolly, Steven R.

AU - Anastasio, Mark A.

AU - Yu, Lifeng

AU - Li, Hua

PY - 2017

Y1 - 2017

N2 - It is widely known that the optimization of imaging systems based on objective, or task-based, measures of image quality via computer-simulation requires use of a stochastic object model (SOM). However, the development of computationally tractable SOMs that can accurately model the statistical variations in anatomy within a specified ensemble of patients remains a challenging task. Because they are established by use of image data corresponding a single patient, previously reported numerical anatomical models lack of the ability to accurately model inter- patient variations in anatomy. In certain applications, however, databases of high-quality volumetric images are available that can facilitate this task. In this work, a novel and tractable methodology for learning a SOM from a set of volumetric training images is developed. The proposed method is based upon geometric attribute distribution (GAD) models, which characterize the inter-structural centroid variations and the intra-structural shape variations of each individual anatomical structure. The GAD models are scalable and deformable, and constrained by their respective principal attribute variations learned from training data. By use of the GAD models, random organ shapes and positions can be generated and integrated to form an anatomical phantom. The randomness in organ shape and position will reflect the variability of anatomy present in the training data. To demonstrate the methodology, a SOM corresponding to the pelvis of an adult male was computed and a corresponding ensemble of phantoms was created. Additionally, computer-simulated X-ray projection images corresponding to the phantoms were computed, from which tomographic images were reconstructed.

AB - It is widely known that the optimization of imaging systems based on objective, or task-based, measures of image quality via computer-simulation requires use of a stochastic object model (SOM). However, the development of computationally tractable SOMs that can accurately model the statistical variations in anatomy within a specified ensemble of patients remains a challenging task. Because they are established by use of image data corresponding a single patient, previously reported numerical anatomical models lack of the ability to accurately model inter- patient variations in anatomy. In certain applications, however, databases of high-quality volumetric images are available that can facilitate this task. In this work, a novel and tractable methodology for learning a SOM from a set of volumetric training images is developed. The proposed method is based upon geometric attribute distribution (GAD) models, which characterize the inter-structural centroid variations and the intra-structural shape variations of each individual anatomical structure. The GAD models are scalable and deformable, and constrained by their respective principal attribute variations learned from training data. By use of the GAD models, random organ shapes and positions can be generated and integrated to form an anatomical phantom. The randomness in organ shape and position will reflect the variability of anatomy present in the training data. To demonstrate the methodology, a SOM corresponding to the pelvis of an adult male was computed and a corresponding ensemble of phantoms was created. Additionally, computer-simulated X-ray projection images corresponding to the phantoms were computed, from which tomographic images were reconstructed.

KW - Image quality assessment

KW - Imaging system optimization

KW - Stochastic object models

UR - http://www.scopus.com/inward/record.url?scp=85020431802&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85020431802&partnerID=8YFLogxK

U2 - 10.1117/12.2254055

DO - 10.1117/12.2254055

M3 - Conference contribution

VL - 10132

BT - Medical Imaging 2017

PB - SPIE

ER -