Karl Baum

Abstract

The benefit of registration and fusion of functional images with anatomical images is well appreciated in the advent of combined positron emission tomography and x-ray computed tomography scanners (PET/CT). This is especially true in breast cancer imaging, where modalities such as high-resolution and dynamic contrast-enhanced magnetic resonance imaging (MRI) and F-18-FDG positron emission tomography (PET) have steadily gained acceptance in addition to x-ray mammography, the primary detection tool. The increased interest in combined PET/MRI images has facilitated the demand for appropriate registration and fusion algorithms. A new approach to MRI-to-PET non-rigid breast image registration was developed and evaluated based on the location of a small number of fiducial skin markers (FSMs) visible in both modalities. The observed FSM displacement vectors between MRI and PET, distributed piecewise linearly over the breast volume, produce a deformed Finite-Element mesh that reasonably approximates non-rigid deformation of the breast tissue between the MRI and PET scans. The method does not require a biomechanical breast tissue model, and is robust and fast. The method was evaluated both qualitatively and quantitatively on patients and a deformable breast phantom. The procedure yields quality images with average target registration error (TRE) below 4 mm. The importance of appropriately jointly displaying (i.e. fusing) the registered images has often been neglected and underestimated. A combined MRI/PET image has the benefits of directly showing the spatial relationships between the two modalities, increasing the sensitivity, specificity, and accuracy of diagnosis. Additional information on morphology and on dynamic behavior of the suspicious lesion can be provided, allowing more accurate lesion localization including mapping of hyper- and hypo-metabolic regions as well as better lesion-boundary definition, improving accuracy when grading the breast cancer and assessing the need for biopsy. Eight promising fusion-for-visualization techniques were evaluated by radiologists from University Hospital, in Syracuse, NY. Preliminary results indicate that the radiologists were better able to perform a series of tasks when reading the fused PET/MRI data sets using color tables generated by a newly developed genetic algorithm, as compared to other commonly used schemes. The lack of a known ground truth hinders the development and evaluation of new algorithms for tasks such as registration and classification. A preliminary mesh-based breast phantom containing 12 distinct tissue classes along with tissue properties necessary for the simulation of dynamic positron emission tomography scans was created. The phantom contains multiple components which can be separately manipulated, utilizing geometric transformations, to represent populations or a single individual being imaged in multiple positions. This phantom will support future multimodal breast imaging work.

Library of Congress Subject Headings

Breast--Imaging; Diagnostic imaging; Tomography; Three-dimensional imaging in medicine; Image processing--Digital techniques

Publication Date

2-1-2008

Document Type

Dissertation

Student Type

Graduate

Department, Program, or Center

Chester F. Carlson Center for Imaging Science (COS)

Advisor

Krol, Andrzej

Advisor/Committee Member

Hornak, Joseph

Advisor/Committee Member

Kerekes, John

Comments

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: RG493.5.D52 B38 2008

Campus

RIT – Main Campus

Download

Share

COinS