Computational Sensing & Imaging Lab, UT Austin
Low-Field MRI Scanner
Feb 2026 — Present
Building a working MRI scanner from scratch — magnets, shimming, and all.
The problem
Clinical MRI machines cost millions of dollars largely because of their superconducting magnets. Low-field MRI trades raw signal strength for radically cheaper, safer hardware — but at low field strength, every microtesla of field inhomogeneity matters. The scanner only produces usable images if the main magnetic field is made extremely uniform across the imaging volume.
What I’m building
I’m constructing a low-field MRI scanner from scratch in the Computational Sensing & Imaging Lab, following the process pioneered at the MRI4ALL hackathon and optimizing the field strength beyond the baseline design. The work spans permanent magnet array assembly, field characterization, and shimming — the parts of MRI that are pure applied electromagnetics.
Field mapping with a hacked 3D printer
To shim a magnet you first have to measure it. Rather than buy a dedicated field-mapping robot, I repurposed a 3D printer into a 3-axis probe positioner: the printer’s motion system scans a magnetometer through the bore on a programmed raster path, producing a dense 3D map of the B₀ field. Those maps directly determine where passive shim material goes and how active shim coils should be driven.
Status
Field mapping and shimming iterations are ongoing, with the goal of a field homogeneous enough for imaging experiments. This project is the center of my interest in medical imaging hardware — the same physics I’m studying in my quantum information coursework, pointed at a machine that can look inside the human body.