VASCTRACE AI

The description states that VASCTRACE AI is a physics-informed PET/CT research copilot. It is described as:

• A research prototype, not a diagnostic system.
• Trained and evaluated on simulated vascular-like FDG sources.
• Built using a 2.5-D shared-weight Siamese U-Net for detecting abnormalities.
• Uses controlled synthetic-source engine to insert parameterized FDG sources into raw SUV data.
• Includes deterministic 3-D quantification, returning SUV statistics or structured nulls with QC reasons.
• Operates on a public QUADRA healthy test/retest cohort (Zenodo 16686025).
• Implements a verifier that rejects generated reports if numbers or claims drift.

The system is described as an end-to-end pipeline, not a product for clinical use, and is explicitly labeled as a research prototype.

Claim: VascuTrace AI is a research tool.

Evidence: The description states it is a "research prototype trained and evaluated on simulated vascular-like sources" and that “no result here establishes clinical sensitivity, specificity, or patient benefit.”

The project’s positioning is rooted in:

• Research methodology — using AI to automate parts of scientific research (planning, architecture, review, writing).
• Reproducibility — emphasizing deterministic code and a verifier that ensures numbers are not altered by generated language.
• Controlled experimentation — focusing on simulated data under controlled conditions rather than clinical diagnostics.

The claim evolution is:

1. Initial claim: A method to detect vascular-like FDG abnormalities in healthy scans using AI.
2. Refinement: The system is a research pipeline, not a diagnostic tool.
3. Clarification: It uses simulated data and is not clinically validated.

Claim: VascuTrace AI is a research copilot for medical imaging.

Evidence: The description states it “owns the parts of research that usually bottleneck a small team” and uses Codex + GPT-5.6 to automate scientific workflows.

The description does not identify any specific customer or target market beyond the project’s own use case as a research tool.

Claim: The product is for researchers in medical imaging.

Inference: Based on the domain (PET/CT, vascular FDG abnormalities), it likely targets those working in radiology or medical imaging research.

Evidence: Not directly stated. The description focuses on the technical pipeline and not on end-users.

There is no evidence of a business model or pricing structure.

Claim: No commercial business model is evident.

Evidence: The project is described as a research prototype, not a product for sale or licensing. No mention of revenue, pricing, or monetization.

The system uses:

• Python 3.13, PyTorch + MONAI, NumPy/SciPy, nibabel/SimpleITK, Streamlit, Pydantic, Model Context Protocol
• Codex + GPT-5.6 for planning, architecture, and report writing
• A Siamese U-Net (B2, deep supervision) for detection
• Deterministic measurement code separate from language generation
• Auditable tool trace, executable evaluation suite, and verifier
• Controlled synthetic-source engine with ground truth by construction

Claim: The system is built with a focus on reproducibility, determinism, and scientific rigor.

Evidence: The description states that “measurement code is physically separate from generated prose” and that “numbers come from pure, side-effect-free functions.”

There is no evidence of traction or commercial adoption.

Claim: No traction or adoption beyond the hackathon project.

Evidence: The project was submitted to a hackathon. No revenue, customers, or usage data are reported.

The description does not mention any direct competitors.

Claim: No competitive landscape is described.

Evidence: No mention of existing tools or systems in the same domain (medical imaging research with AI).

• No clinical validation — The system is explicitly a research prototype, not a diagnostic tool.
• Limited scope — Built for simulated data only; no evidence of real-world deployment.
• Self-reported only — No independent verification or external data to confirm performance claims.
• Hackathon project — Not a commercial product or venture, but a demonstration.

Claim: The system is not ready for clinical use.

Evidence: The description explicitly states that “no result here establishes clinical sensitivity, specificity, or patient benefit.”

1. What are the next steps to move from research prototype to a product or service?
2. Are there any plans to validate this system on real-world data or with clinical partners?
3. Has the team considered how to scale this beyond the current hackathon-level implementation?
4. Is there any interest in licensing or commercializing this research pipeline?
5. What would be required to make this suitable for clinical deployment?