Diffuse Optical Tomography Methods for Imaging the Developing Brain

Diffuse optical imaging (DOI) is a portable imaging modality that provides the ability to perform early and continuous monitoring of brain function in infants. Its methodology overcomes many of the technical and logistical challenges of performing magnetic resonance imaging (MRI) investigations in neonates. However, standard DOI systems suffer from limitations such as low spatial resolution and contamination of hemodynamic signals originating from superficial tissue layers that affect the overall reliability of the optical measurements. Recent advances in the application of high-density diffuse optical tomography (HD-DOT) in adults have overcome most of these limitations by using high-density arrays of overlapping measurements that improve spatial resolution and brain specificity.;My doctoral work has been focused on the development of HD-DOT methods for bedside neuroimaging in neonates. While HD-DOT enables image reconstructions with improved image quality, the design and implementation of high-density arrays for newborn infants involves challenges related to cap ergonomics and bulkiness of the fiber bundles. The first part of my dissertation demonstrates the feasibility of using a custom-built HD-DOT system with a limited field of view for imaging brain function in a clinical environment such as the neonatal intensive care unit (NICU). Using stimulus-driven paradigms such as visual stimulation and advanced imaging techniques such as resting state functional connectivity, I show that HD-DOT can perform functional mapping of the visual cortex in hospitalized infants. Resting state functional connectivity methods are particularly suited for studying hospitalized infants who cannot perform complex tasks. Accordingly, the second part of my dissertation was focused on the development of an extended field of view system for simultaneous functional connectivity DOT (fcDOT) mapping of multiple functional regions. In parallel with the hardware expansion, I developed techniques for realistic forward light modeling and spatial normalization that facilitate anatomical registration between different subjects and imaging modalities. The proposed techniques were validated in vivo against subject-matched functional MRI maps, the gold standard for functional neuroimaging. The strong spatial agreement between individual and group maps obtained for both modalities suggests that fcDOT provides satisfactory spatial localization and resolution for imaging neonates at the bedside. While in most cases it is desirable to use subject-specific structural images for accurate DOT reconstruction, this approach is not sufficient for portable applications. In the last part of my dissertation, I explored the feasibility of using anatomical atlases for forward light modeling. Quantitative comparisons with functional MRI show that atlas-based image reconstruction provides a viable approach to individual head modeling for HD-DOT when anatomical imaging is not available.