The Institute turns noisy measurements into numbers you can defend and act on —
across medical imaging, biostatistics, and signal processing. It takes raw, uncertain data and returns
quantities with honest error bars, so a single number is solid enough to drive a decision. When a
measurement carries the seal, you know what it means and where it came from.
Services
What you can hire it for.
Each engagement turns a measurement into something you can stand behind.
Biomarker calibration & uncertainty
Turn raw imaging parameters (IVIM, ADC, diffusion) into measurements with defensible error bars, so a number can drive a decision.
ML for medical imaging
Denoising, parameter estimation, and identifiability analysis (DnCNN / IVIMnet-class models); know what the data can and can't support.
Survival & biostatistics
Cox/Firth regression, competing risks (IPCW), multiple-testing control, and honest framing of underpowered studies.
Signal & DSP engineering
Real-time multi-engine audio and nonlinear-dynamics synthesis, from research idea to running system.
Research → reproducible artifact
Turn research code into a one-command, seed-fixed pipeline a reviewer can actually run.
Ventures
What it has built.
Systems shipped from the same discipline — rigor first, then something you can run.
Project Cassandra is a PhD-by-publication research program building and validating a calibration layer
for functional-MRI biomarkers in MR-guided radiotherapy — measurements with error bars
trustworthy enough to drive dose decisions.
Detect the shift, but never observe the truth — calibrate and trust a measurement without ever having ground truth.
The arc
Foundation — FashionA calibration ruler for body-IVIM error bars. Several results consume it.
Value of information — MinosWhen a calibrated biomarker is actually worth acting on.
Trust under shift — SibylDetecting out-of-distribution data so you know when to stop trusting the number.
Validation without ground truth — EchoValidating via repeatability data when truth is unavailable.
Capstone aimA closed-loop, calibrated dose-decision system on an MR-Linac.
Ethos
Hope, mission, goal — held apart on purpose.
What drives the work, what it's staked on, and where it's done — kept separate so rigor never rides on luck.
uncontrollable · motivating
Hope
That functional-MRI biomarkers make image-guided radiotherapy improve prognosis.
Allowed to drive the work, never what it's staked on.
controllable · deliverable
Mission
Build and validate the calibration that makes functional-MRI biomarkers trustworthy enough to act on.
Succeeds or fails on rigor and adoption, both in hand; survives a null trial.
concrete · positional
Goal
Direct that work from a research seat in New York.
The seat, plus the calibration contribution.
hope→mission→goal · uncontrollable to controllable to positional
Keep hope and mission separate. If biomarker-adaptive dosing disappoints, the calibration work still stands —
because rigorous, adopted calibration is exactly what lets the field find out. The mission does not depend
on the hope coming true.
Publications
Work in the record.
FashionCalibrated uncertainty for body-IVIM parameters.Magnetic Resonance in MedicineUnder review
AnnealChimera-state machinery for additive synthesis.Nonlinear DynamicsUnder review
MinosValue of information for biomarker-guided dosing.In preparation
SibylOut-of-distribution trust detection under deployment shift.In preparation
EchoGround-truth-free validation via repeatability data.In preparation
Chimera synthesisA self-morphing additive voice from a two-population chimera state.DAFx 2027Targeted