Mapping Microvascular Stress: Our New OSF Project on Stroke-Risk ULM

Intro

Over the last weeks I’ve been working on a slightly unhinged idea:

What if we treated stroke not just as a plaque problem, but as a microvascular fatigue problem we can measure with ultrasound?

Ultrasound localization microscopy (ULM) already lets us see individual microbubbles flowing through tiny vessels. The question I’m exploring is: can we turn those trajectories into stress and fragility maps for the microvasculature – and then wrap that into a realistic clinical workflow?

I’ve now collected the core research notes for this project in an open OSF repository:

OSF project: Vessel Stress Mapping & Stroke-Risk ULM

The blog post below is a guided tour of what’s in there and what I think we’re bringing to the table.

1. Vessel Stress Mapping (VSM-ULM): Geometry of Microvascular Fatigue

Most ULM work today stops at structure and perfusion: velocity maps, vessel density, maybe some dispersion metrics.

In the VSM-ULM notes I push this one step further:

• Treat each microbubble as a Lagrangian stress probe.

• Reconstruct the vessel network and, for each segment, compute a Vessel Stress Index (VSI) from track-level features:

  • axial and oscillatory velocity

  • acceleration peaks

  • residence/dwell time

  • recirculation / direction flips

  • normalized travel distance

On top of that:

• Count “stress cycles” per heartbeat (high-stress vs moderate-stress bins).

• Combine this into a fatigue load per segment.

• Summarize the whole network into a Network Fragility Score (NFS) that emphasizes:

  • high-VSI segments

  • that sit on central paths

  • without good collateral routes.

This is basically importing materials fatigue thinking (S–N curves, cycle counting) into the microvasculature, using ULM tracks as the observable.

OSF file: vsm_ulm_vessel_stress_geometry_research_notes.*

2. Birth-Time Cone Ledger: Physically Honest Tracking

ULM stands or falls on tracking quality. If your tracker randomly stitches bubbles together, you can hallucinate crazy velocities or accelerations and everything downstream is garbage.

The Birth-Time Cone (BTC) Ledger adds a physics + governance layer on top of standard MOT:

• For each bubble birth at time t0​, position x0​, define an admissible cone in space–time based on organ-specific maximum speed and acceleration.

• Any candidate association outside that cone is physically impossible and forbidden.

• Every association decision is written to an append-only ledger with:

  • track ID & birth info

  • which detections were linked

  • BTC margin / violations

  • a quality tier: Clean (A) / Suspicious (B) / Broken (C).

Downstream, VSI/NFS can be computed on:

• A-tracks only (very conservative)

• A + B (sensitive)

• with full coverage / uncertainty maps.

This is aimed at clinical auditability: regulators and clinicians should be able to ask, “Why does this map say this segment is risky?” and we can show the exact tracks and priors behind it.

OSF file: birth_time_cone_ledger_research_notes.*

3. Stress-Reporting Tracer: Painting Where the Walls Suffer

Kinematics alone can tell you where flow is weird, but not necessarily how the endothelium and plaque biology are responding.

The Stress-Reporting Tracer concept adds a second sensing layer:

• Layer A: standard CEUS microbubbles → trajectories, VSM-ULM geometry.

• Layer B: stress-responsive agents designed to:

  • irreversibly change their ultrasound / photoacoustic signal

  • when exposed to pathological conditions such as:

    • high shear

    • plaque-associated enzymes

    • oxidative stress / ROS

    • inflammation at the vessel wall

  • and then stick around long enough to “paint” vulnerable segments.

From this we define:

• SRI_s – Stress Reporter Intensity per vessel segment.

• CRI_s – Composite Risk Index, combining mechanics + biology:

  $${\text{CRI}}_s=\alpha \cdot {\text{VSI}}_s+\beta \cdot {\text{SRI}}_{\mathrm{s<span\; style="font-family:\; Georgia,\; \text{'}Times\; New\; Roman\text{'},\; \text{'}Bitstream\; Charter\text{'},\; Times,\; serif;\; font-size:\; 16px;"></span>}}$$

The notes outline possible chemistries (shear-activated capsules, enzyme-cleavable shells, ROS probes), readout strategies (CEUS vs CEUS+PA), and a validation roadmap.

OSF file: stress_reporting_tracer_research_notes.*

4. CEUS-Only Stroke-Risk Workflow: From Lab Trick to Exam Slot

All of this is useless if it never leaves the lab.

The CEUS Stroke-Risk Workflow notes sketch a realistic exam you could run on a modern ultrasound system:

• Target population: high-risk patients (known plaque, TIA history, heavy risk factors).

• Hardware: existing CEUS-capable scanner + GPU workstation.

• Protocol (carotids):

  • standard duplex first

  • CEUS injection

  • 30–60 s high-frame-rate acquisitions over bifurcations / plaques per side

  • optional stress-tracer sequence

• Processing budget: ~10–15 minutes per patient for:

  • motion correction & clutter filtering

  • ULM + BTC tracking

  • vessel reconstruction + VSI/NFS

  • tracer mapping (if used)

  • risk maps & report export.

Outputs are designed to be boring in a good way:

• Color-coded vessel maps (green–yellow–red)

• NFS per territory

• Top-N high-risk segments with short descriptions

• Left–right asymmetry indices

• Coverage / QA maps showing where the data is strong vs weak.

OSF file: ceus_stroke_risk_workflow_research_notes.*

5. System-Level Framing: Stroke as Terrain + Fatigue

The last note is the glue.

Instead of treating plaques as isolated villains, I frame the whole thing as:

A coupled macro–microvascular system under long-term mechanical and biochemical load.

In that language:

• VSM-ULM → measures how the load is distributed across the network.

• BTC ledger → makes sure those measurements obey physics and are traceable.

• Stress tracer → shows where the vessel wall is actually suffering.

• CEUS workflow → turns it into a repeatable sensor you can run every year.

The goal is a microvascular risk OS, not just a prettier Doppler image.

OSF file: system_level_microvascular_risk_framing_research_notes.*

Why OSF, and What Happens Next?

I’m putting these notes on OSF for a few reasons:

• To make it easy for ultrasound / ULM groups, clinicians, and nanomedicine folks to plug in.

• To keep a versioned, citable record as the ideas evolve.

• To lower the barrier for people who want to sanity-check, attack, or extend any part of the stack.

Again, the project is here:
https://osf.io/25grf

If you work on:

• ULM / sULM algorithms

• CEUS clinical workflows

• Shear- or enzyme-responsive contrast agents

• Or you’re just very into pipes, fatigue and stroke prevention

…then have a look, poke holes in it, and feel free to reach out.