PhD Thesis Proposal: Mengyang Zhao

Optional ZOOM LINK

"3D MRI-guided Near-infrared Spectroscopic Tomography for Imaging Breast Cancer"

Abstract

MRI-guided Near-infrared Spectroscopic Tomographic (MRg-NIRST) imaging, a non-invasive and non-radioactive imaging tool, has shown potential advantages for breast cancer detection. In this study, a novel MRg-NIRST imaging system with an MRI-compatible and flexible breast optical interface has been developed to provide tumor functional information about physiologically related biomarkers such as oxy-hemoglobin, deoxy-hemoglobin, and water. This new interface is equipped with 48 photodetectors (PDs) and 48 firing fibers, designed to accommodate breasts of various sizes and shapes. This configuration allows data acquisition from up to 2304 source-detector positions concurrently with MRI scans, enabling 3D MRg-NIRST reconstruction of the entire breast. The optical data at six wavelengths over the range of 660nm to 852 nm can be collected simultaneously with MRI scanning in around 4 minutes. A patient-specific meshing procedure was developed to achieve a 3D mesh based on the co-registered MRI images for finite element method (FEM) reconstruction. The system has been validated through a series of phantom studies and normal subject exams. Reconstructed images of heterogeneous phantoms showed sharp contrasts in inclusions relative to the background, with accurately recovered inclusion sizes. Total hemoglobin (HbT) concentration values estimated from the images of normal subjects were consistent with those obtained in previous imaging studies.

Although the current system and reconstruction approach have been validated through phantom and normal subject studies, several critical aspects remain to be addressed in the final phase of this thesis. First, an upcoming patient study will focus on identifying key optical biomarkers and comparing the diagnostic power of MRg-NIRST with that of MRI alone for breast cancer detection. Second, efforts will be made to reduce surface artifacts in MRI images by optimizing the flex circuit design and incorporating new strips with minimal metal elements. Finally, systematic optimization of the current calibration and image reconstruction workflow will be conducted including the development of a calibration procedure for complex 3D mesh geometry to further suppress surface artifacts.

Thesis Committee

• Shudong Jiang (Co-chair)
• Keith Paulsen (Co-chair)
• Petr Bruza
• Darren Roblyer (Boston University)

Contact

For more information, contact Thayer Registrar at thayer.registrar@dartmouth.edu.