Autonomous Radiology Intelligence
VoxelMD builds clinical-grade AI that reads medical images and drafts diagnostic reports — targeting the massive, underserved radiology automation market. Self-funded by a practicing Neuroradiologist who both codes and practices medicine.
The Crisis in Radiology
Demand for imaging is exploding while the workforce cannot keep up. This is a systemic, structural problem — and it's getting worse.
The American College of Radiology reports critical workforce shortages across all subspecialties.
Scan volumes grow relentlessly every year, outpacing the number of trained radiologists entering the workforce.
A significant portion of a radiologist's day is consumed by manual dictation rather than diagnostic interpretation.
Our Solution: Two Complementary Products
VoxelMD attacks the problem from two angles — automation of tedious tasks today, and autonomous image interpretation tomorrow.
Voxel Suite
- Reduces dictation time by up to 80% via intelligent clinical data injection
- Summarizes 5 years of prior imaging findings at point of care
- Integrates seamlessly with Epic and enterprise PACS
- 10 active beta radiologists, ~30% throughput increase per shift
Voxel Vision
- AI that reads DICOM images and drafts full diagnostic reports
- Dual deployment: on-premise NVIDIA GPU (Gemma 4 / MedGemma, Qwen 3.5 VL) or HIPAA cloud (Vertex AI + Gemini with BAA)
- Thyroid Ultrasound reader and CT Perfusion drafter prototypes completed
- Follows the DICOM-in/results-out architecture used by RapidAI and Rad AI
- Long-term FDA pathway to direct CPT code billing ($100–$900+/study)
Market Opportunity
The radiology AI market is early, massive, and growing at 15–25% CAGR. We are positioned at the intersection of clinical credibility and technical execution.
AI in Healthcare
Radiology AI Market
Private Practice Automation
Sources: Grand View Research, MarketsandMarkets, Mordor Intelligence, Precedence Research (2025–2026 reports)
The Unfair Advantage
Mo Fakhri, MD — Practicing Neuroradiologist, sole founder, and the person writing every line of code. Unlike most health-tech startups built with a tech-only orientation, VoxelMD is uniquely driven by a medical founder who deeply understands the everyday pain points, clinical needs, and high-stakes questions of practicing radiologists.
Trained at Harvard Medical School (research fellow), UCSF (neuroradiology fellowship), and Mallinckrodt Institute of Radiology (residency). Recipient of the NIH T32 Research Grant and RSNA Research Resident Grant.
Full-Stack Capability: Architected a dual deployment system — on-premise NVIDIA GPU inference for air-gapped hospitals, and HIPAA-compliant cloud inference via BAA-covered Vertex AI for cost-efficient scaling.
Self-funded from day one. Every dollar invested to date comes from the founder — demonstrating full personal conviction in the mission and technology.
Revenue Model
A clear, two-track path to revenue with compounding returns.
Track 1: B2B SaaS Licensing
Sell Voxel Suite directly to private radiology practices and hospitals. Per-seat or per-practice licensing with enterprise tiers for departments.
Track 2: Direct CPT Code Billing
Pursue FDA clearance for Voxel Vision modules. Map AI-generated interpretations to billable CPT codes, transforming AI from practice overhead into an independent revenue center billed directly to insurance.
Validation & Proof of Concept
Selected for high global volume and critical diagnostic impact — proving our architecture before scaling.
Thyroid Ultrasound
CPT 76536 · $100–$125 / studyUp to 50% of adults have thyroid nodules, generating massive imaging volume globally ($800M+ market). Our prototype successfully matches radiologist accuracy in TI-RADS classification.
CT Perfusion (Stroke)
Time-Critical · Acute CareOur vision models ingest complex color-coded perfusion maps, interpret core/penumbra mismatches, and draft comprehensive dictations — proving capability to read complex multi-series inputs under high-stakes constraints.
Uncompromising HIPAA Compliance
We offer an Enterprise On-Premise Option — our models can run entirely on local hardware. This air-gapped architecture ensures Protected Health Information (PHI) never traverses the public internet, satisfying the most rigid hospital IT and cybersecurity requirements.
Why Now
Multimodal AI models have reached clinical-grade accuracy for the first time
Global radiologist shortage is accelerating — demand outpaces supply structurally
Regulatory environment (FDA) increasingly receptive to AI-assisted diagnostics
Healthcare systems actively seeking automation to control costs and reduce burnout
Partner With Us
We are seeking pre-seed investment to accelerate product development, expand our beta program, and prepare for FDA pre-submission.
Return to Home