Optimization of Fast MR Imaging Technologies using the Case-PDM to Quantitatively Assess Image Quality

There exist an extraordinary number of ways to create an MR image, and the number seems to grow daily. In almost all cases, images will be viewed by radiologists to make a diagnosis, to stage a disease, to apply a treatment, and/or to assess a treatment. Since images are viewed, one needs to assess visual image quality, preferably in a quantitative way. We developed a perceptual difference model, Case-PDM, to quantitatively assess image quality, in a way well-correlated to clinical needs, and demonstrate how the methods can be used to improve MR imaging techniques. We validated existing Case-PDM with advanced observer experiments. Human evaluation of MR images from multiple organs and from multiple image reconstruction algorithms were compared to Case-PDM and competing methods such as IDM (Sarnoff Corporation) and SSIM (Wang et al.). Global image quality is quantified by comparing fast acquired, reconstructed image to slower, full k-space, high quality reference image. We used advanced human observer experimental methods (DSCQS, FMT, 2AFC) to prove these methods on the contexts of human-model correlation, comparability of model evaluation scores across different image contents, and imperceptible difference threshold discrimination. To date, most objective image quality metrics average over a wide range of image degradations. However, human clinicians demonstrate bias toward different types of artifacts. We used an advanced observer experiment and Artifact-PDM, an extension of Case-PDM, to measure relative disturbance of MR image artifacts to radiologists. We used a Functional Measurement Theory (FMT) pair-comparison experiment to measure the disturbance of each artifact to human observers. Radiologists showed preferences towards particular image artifacts, the relative disturbances of which can be quantitatively measured by both observer study and Artifact-PDM. We also applied our methodology to a novel MRI reconstruction algorithm, such as Compressed Sensing (CS) which allows MRI reconstruction from incoherent/random partial k-space samples, to optimize various parameters and study parametric sensitivity, using a large number in vivo MRI data. We conclude that our PDM can faithfully represent human observer image quality evaluation and can be useful in evaluating reconstruction algorithms, especially in evaluating artifact trade-offs, to improve diagnostic accuracy for clinical protocols.

Name thesis:degree_name

Doctor of Philosophy

Level thesis:degree_level

doctoral

Discipline thesis:degree_discipline

Biomedical Engineering

Grantor dc:publisher

Case Western Reserve University School of Graduate Studies

Year dc:date

2013

Author dc:creator

• Miao, Jun

Contributors dc:contributor

• Wilson, David

• Biomedical Engineering
• Medical Imaging
• Perceptual difference model
• Image quality
• Image artifact
• Magnetic Resonance Imaging
• Reconstruction Algorithm
• Observer Study

Statement dc:rights

• unrestricted
• This thesis or dissertation is protected by copyright: all rights reserved. It may not be copied or redistributed beyond the terms of applicable copyright laws.

Language dc:language

English

Repository record dc:identifier

http://rave.ohiolink.edu/etdc/view?acc_num=case1346966179

OAI identifier oai:identifier

oai:etd.ohiolink.edu:case1346966179