No cell in your body is more than a few hundreds of microns from a blood vessel. A key challenge for artificial organs is developing a biomimetic vasculature to support cell growth. In this work I used the large three-photon absorption cross section of silk fibroin to generate 3D structures in silk hydrogels that range in size from a few microns up to nearly a millimeter at unprecedented depths

Photocrosslinking silk with riboflavin

Worked on in: 2016 – (Publication link)

Silk is a great biocompatible, transparent substrate for all sorts of biomedical applications. One such is to use silk to patch corneal wounds. I developed a method to transform a liquid silk solution into a hydrogel when exposed to blue or UV light that binds readily to corneal tissue. A provisional patent on this method has been filed (62/268,993).

Multi-distance diffuse optical spectroscopy with a single optode via hypotrochoidal scanning

Worked on in: 2018 – (Publication link)

In diffuse optical spectroscopy, the distance between the source and detector fiber determines how deep below the surface detected light has traveled. Longer separation is more sensitive to deeper structures, while short separation is more sensitive to superficial changes. Many instruments rely on an array of different sources and detectors to build up 3D estimates of tissue hemoglobin. Here, I use mechanical scanning to trace the source and detector through a hypotrochoid which allows for a range of source/detector separations that are all centered at the same location. This allows measurements to form a pseudo axial line scan to build up 2- and 3-dimensional images of tissue chromophores.

OpenSFDI

Worked on in: 2020 – (Publication link)

Spatial Frequency Domain Imaging is a diffuse optical imaging method able to quantify the concentration fo hemoglobin in tissue. OpenSFDI is a wesite that provides step-by-step instructions for building an SFDI system.

Real-time diffuse optical spectroscopy

Worked on in: 2021 – (Publication link)

Diffuse optical spectroscopy has a long history of use for the monitoring of breast cancer, but lengthy acquisition and processing times have made it difficult to use in the clinic. While in the Roblyer lab, I helped develop an ultrafast spectroscopy system with onboard data processing and integrated it with a novel probe tracking technique to provide real-time information of tissue hemodynamics.

Machine learning for imaging skin cancer with reflectance confocal microscopy

Worked on in: 2022

Reflectance confocal microscopy (RCM) is a noninvasive optical imaging modality capable of achieving cellular resolution that is currently used to rapidly diagnose skin cancer. I’m working on developing machine learning algorithms that will make it possible for the next generation of RCM microscopes smaller, faster, and cheaper.

Self-supervised denoising of volumetric biomedical images

Worked on in: 2023 – (Publication link)

Improving medical images with machine learning is hard because of the difficulty in collecting datasets that contain gold standard, or ground truth images. Self-supervised methods can learn to denoise images solely from noisy data. In this work, I show how similarities between adjacent images in a volume can be used to remove noise from a variety of volumetric biomedical image data.