# Generative Dose Control Calibration

*/Opportunities/Generative_Dose_Control_Calibration*

## Opportunity Overview

**Wedge**: Target independent outpatient radiation clinics treating prostate cancer as the initial beachhead. Prostate plans possess standard, well-defined constraint parameters for sparing the rectum and bladder and constitute a massive volume of daily clinic workload, enabling rapid proof of value. Expand subsequently into complex head-and-neck cancer treatment plans, followed by integration into diagnostic CT dose reduction calibration.
**Timing**: Recent advances in 3D generative vision models enable accurate predictions of achievable dose distributions directly from volumetric patient anatomy scans. Simultaneously, new API access to major treatment planning systems allows direct, automated parameter adjustment without manual data entry.
**Why This I C P**: Radiation oncology clinics operate under strict throughput bottlenecks driven by a severe, systemic shortage of certified medical physicists. They also possess clear, quantifiable success metrics via Dose-Volume Histograms that make objective evaluation of an automated tool immediate and indisputable.
**Size Of Prize**: There are approximately 9,000 radiation oncology centers globally that each spend an average of $50,000 annually in medical physicist labor time specifically on iterative dose tuning. Multiplying these 9,000 centers by the $50,000 labor offset yields a $450M addressable prize.
**Gap Narrative**: Medical physicists currently spend hours manually tuning dose volume constraints and calculating plans for radiation therapy machines to meet clinical goals. Existing treatment planning systems require iterative, trial-and-error manual adjustments to spare healthy organs while delivering lethal doses to tumors. Generative Dose Control Calibration produces optimized machine configurations instantly by predicting the optimal 3D dose distribution and back-calculating the required multi-leaf collimator parameters.
**Defensibility**: Defensibility compounds through a proprietary dataset of accepted versus rejected dose-volume plans, creating a workflow lock-in as the model fine-tunes to a specific clinic's unique physician preferences. The tight integration into the clinic's rigid, FDA-regulated treatment planning system establishes extreme switching costs, as replacing the calibration engine requires extensive recommissioning and clinic downtime.
**Why This Thesis**: A Service-as-Software approach directly ingests patient scans and physician prescriptions to output a fully calibrated machine plan. This offloads the entire iterative tuning task rather than just providing another software interface, directly substituting for the missing physicist labor hours.

## Opportunity Linked Thesis

**Thesis**: [Software](/Theses/Software)

## Opportunity Linked I C P

**Icp**: [Radiation Oncology Center](/CompanyTypes/Radiation_Oncology_Center)

## Opportunity Market Sizing

_Illustrative — target and order-of-magnitude estimate figures, not an achieved track record (this Thing is concept-stage)._

**S A M**: ~$200-300M US and Western European regional health systems operating modern high-throughput linear accelerators
**S O M**: ~$10-25M realistic 3-year capture targeting independent US oncology networks
**T A M**: ~12,000 global radiation oncology centers × ~$60,000/yr software and physicist labor displacement ≈ ~$720M
**Growth Rate**: ~10-15%/yr driven by the transition to hyper-fractionated therapies like SBRT requiring denser dosimetric QA overhead
**Paid Comparable Spend**: ~$80,000-120,000/yr per center in medical physicist labor for beam modeling and legacy phantom hardware QA

## Opportunity Incumbents

- [Varian Eclipse](/Products/Varian_Eclipse) — Tool
- [RaySearch RayStation](/Products/RaySearch_RayStation) — Tool
- [Elekta Monaco](/Products/Elekta_Monaco) — Tool
- [Custom Python Scripts](/Products/Custom_Python_Scripts) — DIY
- [Manual Excel Worksheets](/Products/Manual_Excel_Worksheets) — Spreadsheet
- [Outsourced Physics Consultants](/Products/Outsourced_Physics_Consultants) — Service
- [OpenGATE Simulation](/Products/OpenGATE_Simulation) — Open-Source

## Opportunity Win Conditions

**Kill Thresholds**:
- Manual adjustment rate exceeds 25 percent after 60 days
- Time-to-generate exceeds 60 minutes per linear accelerator
- Trial conversion rate falls below 20 percent in the first 90 days
- More than 2 consecutive clinic compliance reviews block deployment
- Human verification consumes more than 2 hours per calibration plan
**Leading Metrics**:
- Time-to-first-generated-beam-model
- Manual adjustment rate per generated calibration plan
- Weekly active physicists executing dosimetric QA checks
- Trial-to-paid conversion rate
- Human-in-the-loop review minutes per plan
**What Proves Right**: Medical physicists successfully replace manual phantom hardware QA and beam modeling with the generative calibration output. Clinics convert to a $4,000/month subscription after a 30-day trial because the software cuts per-machine calibration time from 6 hours to under 30 minutes. Cohort retention remains above 90 percent at month six as generated models consistently pass independent clinical validation tests without manual adjustments.
**What Proves Wrong**: Medical physicists refuse to trust the generated dosimetric outputs and run parallel manual phantom QA indefinitely. Compliance officers block clinical deployment because the model fails to explicitly map underlying Monte Carlo simulation constraints. The time saved is consumed by human-in-the-loop verification, resulting in no net labor reduction or cost savings for the oncology center.

## Opportunity Build Profile

**Hardest Part**: Guaranteeing the physics constraints of the generative model so it never recommends a dose parameter that degrades image quality below diagnostic utility or exceeds clinical safety thresholds.
**Min Viable Scope**: V1 acts purely as an offline recommendation engine for non-contrast head CT protocols on a single scanner manufacturer, outputting suggested adjustments for the physicist to review. Exclude direct integrations that push protocol updates to the scanner automatically and ignore complex multi-phase scans.
**Cold Start Problem**: The models require paired datasets of patient habitus, dose parameters, and resulting diagnostic image quality, which are locked in hospital PACS and dose monitoring silos. Break this by partnering with a single academic medical physics department to train on retrospective Monte Carlo simulations using historical phantom data.
**Time To First Value**: 3-6 months of local phantom testing and medical physics validation
**Data Moat Available**: true
**Technical Difficulty**: Very High

## Neighborhood

### Surfaced from

- [Lithography System Manufacturers](/CompanyTypes/Lithography_System_Manufacturers) — surfaces · CompanyTypes

### Incumbent in

- [Bespoke Python Scripts](/Products/Bespoke_Python_Scripts) — incumbent in · Products
- [Varian Eclipse](/Products/Varian_Eclipse) — incumbent in · Products
- [Outsourced Physics Consultants](/Products/Outsourced_Physics_Consultants) — incumbent in · Products
- [RaySearch RayStation](/Products/RaySearch_RayStation) — incumbent in · Products
- [Elekta Monaco](/Products/Elekta_Monaco) — incumbent in · Products
- [Manual Excel Worksheets](/Products/Manual_Excel_Worksheets) — incumbent in · Products
- [OpenGATE Simulation](/Products/OpenGATE_Simulation) — incumbent in · Products

### Applies thesis

- [Radiation Oncology Center](/CompanyTypes/Radiation_Oncology_Center) — applies thesis · CompanyTypes

### Embodies

- [Software](/Theses/Software) — embodies · Theses

### Similar Opportunities

- [NRC Filing Service](/Opportunities/NRC_Filing_Service) — similar · Opportunities
- [Autonomous Claim Adjudicator](/Opportunities/Autonomous_Claim_Adjudicator) — similar · Opportunities
- [Claim Resolution Engine](/Occupations/Healthcare_Practitioners_and_Technical_Occupations/Opportunities/Claim_Resolution_Engine) — similar · Opportunities
- [IRB Proxy](/Opportunities/IRB_Proxy) — similar · Opportunities
- [Equipment Tuning Service](/Opportunities/Equipment_Tuning_Service) — similar · Opportunities
- [Synthetic Mock Site Visits](/Opportunities/Synthetic_Mock_Site_Visits) — similar · Opportunities
- [Automated Prior Authorization](/Opportunities/Automated_Prior_Authorization) — similar · Opportunities
- [Pathology Triage Engine](/Opportunities/Pathology_Triage_Engine) — similar · Opportunities
- [Autonomous Calibration for Water Plants](/Opportunities/Autonomous_Calibration_for_Water_Plants) — similar · Opportunities
- [Autonomous Dossier Structuring for Pharma](/Opportunities/Autonomous_Dossier_Structuring_for_Pharma) — similar · Opportunities
- [AI Medical Claims Adjudication for Payers](/Opportunities/AI_Medical_Claims_Adjudication_for_Payers) — similar · Opportunities
- [PID Tuning API](/Opportunities/PID_Tuning_API) — similar · Opportunities
- [Infusion Yield Optimization](/Opportunities/Infusion_Yield_Optimization) — similar · Opportunities
- [Autonomous Dossier Compilation for Hardware](/Opportunities/Autonomous_Dossier_Compilation_for_Hardware) — similar · Opportunities
- [Regulatory Standard Mapping for MedTech](/Opportunities/Regulatory_Standard_Mapping_for_MedTech) — similar · Opportunities
- [Patient Retention Agent](/Opportunities/Patient_Retention_Agent) — similar · Opportunities
- [Autonomous Scan Triage for Level III](/Opportunities/Autonomous_Scan_Triage_for_Level_III) — similar · Opportunities
- [Hospital Clinical Record Parsing](/Opportunities/Hospital_Clinical_Record_Parsing) — similar · Opportunities
- [AI Repair Diagnostics for Dealerships](/Opportunities/AI_Repair_Diagnostics_for_Dealerships) — similar · Opportunities
- [Intelligent Bed Management for Hospitals](/Opportunities/Intelligent_Bed_Management_for_Hospitals) — similar · Opportunities