Market Snapshot & Outlook

The global CAD mammography market cleared $1.10 billion in 2025 and is projected to reach $2.70 billion by 2035, a 9.40% compound annual growth rate, according to Precedence Research. The adjacent breast imaging software category — the broader bucket that includes CAD, workflow, and reporting tools — was valued at approximately $479 million in 2025, growing at a 7.5% CAGR through 2033 per Data Insights Market. The 3D (tomosynthesis) segment led the CAD mammography category with roughly 55% share in 2025, reflecting the continued shift away from 2D-only screening. For a founder writing a business plan, the useful signal in these numbers isn't the headline market size - it's the growth rate and the segment mix, both of which point the same direction: capital committed to 3D-capable, AI-CAD-ready equipment ages better than capital committed to a 2D-only, legacy-CAD setup.

The category is being rebuilt around AI, not the old CAD vendors

The most important thing a 2026 business plan needs to get right is that "CAD" no longer means what it meant a decade ago. Roughly 74% of US screening mammograms were read with legacy, rules-based CAD assistance about thirteen years ago. Since then, deep-learning AI-CAD has overtaken the category: peer-reviewed studies now associate AI-based systems with materially fewer false positives and faster reading times than the older generation of software, and every serious new contract in this space is being written around an AI vendor, not a legacy CAD box.

The vendor landscape splits roughly three ways. Hologic holds the largest single share of the AI-enhanced mammography screening market at 18.7% (2024), and dominates hospital-chain contracts alongside GE HealthCare and Siemens Healthineers. iCAD has focused on penetrating outpatient imaging centres rather than hospital systems. ScreenPoint Medical, maker of the Transpara platform, has grown through partnerships with regional teleradiology groups offering cloud-based AI reading. Volpara Health Technologies and Three Palm Software (maker of the CadOne platform) round out a longer list of specialist players that also includes Lunit, Kheiron Medical Technologies, Therapixel, and Fujifilm's imaging division. A new practice's plan should explicitly state which of these it is standardising on and why — reviewers can tell within a paragraph whether a plan understands the category or is copying 2015-era assumptions.

Two policy shifts are reshaping demand on both sides of the Atlantic. In the US, the 2024 USPSTF update lowered the recommended starting age for biennial screening from 50 to 40, materially expanding the addressable screening population a new practice can serve. In the UK, the NHS launched what has been described as the world's largest AI-led breast cancer screening trial in early 2025: roughly 462,000 of 700,000 screening studies across 30 centres are being double-read by five competing AI systems, testing whether software can safely replace one of the two human readers currently required per mammogram. Early European data cited alongside the trial suggest a 15-20% gain in cancer detection without a corresponding rise in false positives — the kind of number that should anchor the clinical-quality section of any UK-facing plan.

For a business plan, the practical takeaway is to name the specific AI-CAD platform the practice will run on, cite its FDA clearance or UKCA/CE marking status, and tie the choice to a concrete reimbursement and workflow story rather than a generic "we use advanced technology" claim. Related guides worth reading alongside this one: our diagnostic imaging business plan template and radiology center business plan template, both of which share the same reimbursement and accreditation mechanics.

Four business models hiding under one keyword

"CAD mammography business" gets used loosely, and the plan needs to be explicit about which of four quite different models is actually being built, because the capital stack, licensing burden, and revenue timeline differ sharply between them.

• Independent breast imaging centre: owns the mammography unit, employs or contracts technologists, and either employs an interpreting radiologist or contracts one on a per-read basis. This is the model this page is built around, because it carries the full $185K-$620K capital stack and the full MQSA/CQC accreditation burden.
• AI-CAD add-on to an existing practice: a radiology group or hospital outpatient department that already has mammography units simply licenses an AI-CAD platform. Capital outlay is a fraction of the standalone-centre figure - typically the $20K-$55K software/integration line above, with no new equipment or accreditation needed.
• Teleradiology AI-reading service: a remote reading business that partners with imaging centres to provide overflow or second-opinion reads, layering AI-CAD triage on top of a distributed radiologist panel. Capital needs are software- and staffing-led rather than equipment-led, but state-by-state licensure for remote-reading radiologists becomes the binding regulatory constraint.
• Mobile mammography unit: a truck- or trailer-mounted mammography suite serving underserved or rural populations on a rotating schedule. Equipment cost is similar to a fixed site, but the vehicle, routing logistics, and state-by-state mobile-facility permitting add a distinct layer of complexity not covered by a standard fixed-site plan.

Most lenders and investors reading a "cad mammography" plan will assume the first model unless told otherwise, so the executive summary should state the model explicitly in its first paragraph.

Referral channels that actually drive volume

An independent breast imaging centre rarely wins patients through direct-to-consumer marketing alone; the plan should map three referral layers explicitly. Primary care and OB-GYN networks generate the largest share of routine screening referrals and are won through practice-liaison relationships and fast turnaround on reports, not advertising spend. Hospital and health-system overflow is the second layer - many hospital-based breast centres run at or beyond capacity and will contract out screening volume to an independent site that can turn around next-day appointments, particularly once the 2024 USPSTF age change pushed more patients into the screening pool. Patient self-referral and reminder systems - annual recall notices, employer wellness programmes, and community health outreach - fill the remaining volume, and are the channel most within a small practice's direct control. A plan that names specific referring practices or health systems by category (not necessarily by name, for confidentiality) and states a target referral mix by year two reads as materially more credible to a lender than one that describes marketing in generic terms.

Common Questions Before You Build the Plan

These are the questions investors, lenders, and first-time operators ask most often before committing capital to a CAD or AI-mammography venture. Answering them clearly inside the plan itself - rather than leaving a lender or investor to ask them in a follow-up call - removes the single biggest source of diligence delay, and signals that the founder understands the category as it exists in 2026 rather than as it was described in an older, generic healthcare template.

What is CAD in mammography and how does it work?

Computer-aided detection is software applied to a digital mammogram after acquisition. It marks candidate regions — clusters of microcalcifications or masses — for the radiologist to review; it does not issue a diagnosis on its own. The first commercial CAD venture in mammography dates to 1993, and the first FDA-cleared unit followed in June 1998. Because CAD is applied to images post-acquisition rather than controlling image quality at capture, the FDA has historically treated it as sitting largely outside direct MQSA equipment regulation, even though the facility performing the underlying mammogram must still be MQSA-accredited.

Is CAD mammography still used, or has it been replaced by AI?

Legacy, rules-based CAD is being actively displaced. Adoption of the older technology once reached roughly 74% of US screening studies; today, growth in the category is concentrated almost entirely in deep-learning AI-CAD platforms, which clinical literature associates with fewer false positives and better detection sensitivity. Any plan still describing "CAD" as a single static technology, rather than naming an AI-CAD vendor, will read as dated to a lender or investor who has looked at this space recently.

Does insurance actually cover CAD mammography as a line item?

Not separately, since 2017. CAD used to have its own CPT add-on code that could be billed alongside a base mammography code. The 2017 CPT overhaul folded CAD into the base codes themselves — 77065 (diagnostic, unilateral), 77066 (diagnostic, bilateral) and 77067 (screening, bilateral) — so it is reimbursed as part of the standard exam. A plan that models CAD as a stand-alone billable line, rather than bundled reimbursement, is working from a pre-2017 assumption.

What's the realistic timeline from decision to first billable exam?

Assume 4-7 months end to end for a single-site launch: 60-120 days for MQSA/ACR facility accreditation once equipment is installed, plus procurement lead time on the imaging unit (8-16 weeks for a new 3D system), plus credentialing time for interpreting radiologists with each payer network. Plans that compress this into "launch in 60 days" are the ones that lose credibility with lenders fastest.

Do I need my own AI-CAD algorithm, or can I license one?

Almost no new independent practice builds its own algorithm. The standard path is licensing a cleared platform — Transpara, iCAD's ProFound AI, Volpara, or a similar vendor — on a per-exam or subscription basis. Building proprietary AI is a venture-scale undertaking measured in tens of millions of dollars and multi-year clinical validation; it belongs in a plan only if the business genuinely is an AI vendor, not a screening provider.

How do I choose between the AI-CAD vendors?

Weigh four things in the plan rather than picking on brand recognition alone: whether the platform is cleared for both 2D and 3D (tomosynthesis) studies, whether pricing is per-exam or an annual subscription (per-exam suits lower initial volume; subscription suits a practice confident in steady throughput), whether the vendor already integrates with the PACS/RIS system chosen for the site, and whether the vendor has published peer-reviewed validation on a population similar to the practice's expected patient mix. Lenders respond well to a plan that names two shortlisted vendors and states the selection criteria, rather than a single unexplained brand choice.

Startup Costs & Capital Stack

Launching a CAD or AI-enabled mammography practice typically requires $185K to $620K (roughly £146K to £490K) in initial capital. The spread is driven almost entirely by one decision: refurbished 2D equipment versus a new 3D tomosynthesis system, and how much working capital is set aside before revenue ramps.

Cost Breakdown

• Mammography imaging unit (refurbished 2D FFDM to new 3D tomosynthesis): $75K–$260K (£59K–£205K)
• AI-CAD software licensing + PACS/workstation integration: $20K–$55K (£16K–£43K)
• Facility build-out & radiation shielding: $35K–$90K (£28K–£71K)
• ACR/MQSA accreditation, licensing & compliance: $8K–$28K (£6K–£22K)
• Staff recruitment (technologists + interpreting radiologist contracts): $30K–$95K (£24K–£75K)
• Working capital reserve (approx. 6 months): $17K–$92K (£13K–£73K)

On equipment specifically: a refurbished 2D digital mammography unit with a warranty and service contract runs roughly $40K-$120K, while a new full-field digital mammography (FFDM) system runs $150K-$250K and a new 3D tomosynthesis system runs $250K-$500K depending on brand and configuration, per Block Imaging's 2026 pricing guide. Radiologist workstations add $15K-$25K on the secondary market, and detector paddles run $700-$2,000 each. Given the market's 55% tilt toward 3D tomosynthesis, most lenders now expect a new practice to justify a 2D-only purchase explicitly rather than assume it as the default - a refurbished 3D unit at the upper end of the refurbished-price range is increasingly the more defensible middle path between cost and future-proofing.

Funding Routes

In the US, SBA 7(a) loans are the dominant funding route for this category — amounts run $350,001 to $5 million, with 7-10 year terms for equipment and working capital. Equipment financing companies and industry-specific grants supplement bank financing. In the UK, Start Up Loans (up to £25,000 at a 6% fixed rate) rarely cover the full capital stack on their own for equipment this expensive, so most UK founders pair a Start Up Loan for working capital with a commercial equipment-leasing facility or asset-finance loan secured against the mammography unit itself. Many founders on both sides of the Atlantic also combine personal savings, physician-partner buy-ins, and vendor-financed equipment leases to reduce the cash needed at close.

A phased launch reduces the capital ask

Lenders respond well to a plan that sequences capital rather than asking for the full $620K on day one. A common structure: raise for a refurbished 2D or 3D unit and a per-exam AI-CAD subscription first (the lower end of the range), prove out referral volume and payer credentialing for 12-18 months, then raise a second, smaller tranche to upgrade to a new tomosynthesis system or add a second suite once utilisation data exists. This staged approach also directly addresses the accreditation-timeline risk lenders flag most often — a phase-one plan only needs to clear MQSA/ACR accreditation for one modality before it starts generating the cash flow that supports phase two.

Ongoing costs matter as much as the launch number. Maintenance contracts on the imaging unit typically run $10,000-$20,000 a year per unit, and ACR accreditation must be renewed periodically rather than being a one-time cost — both lines belong in the five-year forecast, not just the opening balance sheet.

SBA & Equipment-Financing Data

Diagnostic imaging businesses sit under NAICS 621512 (Diagnostic Imaging Centers), which carries an SBA small-business size standard of $19 million in average annual receipts — meaning almost every independent CAD-mammography startup qualifies as a small business for SBA purposes.

Healthcare businesses generally outperform the broader small-business population on approval rates — 60-65% at traditional banks, above the all-industry average — and the sector accounts for roughly 8-10% of total annual SBA 7(a) loan volume. Lenders underwriting a CAD-mammography application will want to see three things a generic healthcare plan often omits: the specific AI-CAD platform and its cost per exam, the MQSA/ACR accreditation timeline built into the launch schedule, and a payer-mix assumption that reflects the bundled CPT codes rather than a stand-alone CAD reimbursement line. Plans that get these three details right move noticeably faster through underwriting.

Where the numbers come from matters to an underwriter as much as the numbers themselves. Average medical-equipment transaction size across all specialties rose from roughly $71,000 in 2022 to $87,000 in 2025 — a data point worth citing directly if the plan's equipment line sits above that figure, since it shows the founder understands the current pricing environment rather than quoting a stale number from an older template. Equipment loans in this category are typically structured with the mammography unit itself as collateral, which is one reason approval rates for healthcare equipment financing run ahead of unsecured small-business lending more broadly.

Revenue Model & Reimbursement Economics

Revenue in this category is reimbursement-driven, not price-set. The three CPT codes that matter are 77065 (diagnostic mammography, unilateral, including CAD), 77066 (diagnostic, bilateral, including CAD), and 77067 (screening mammography, bilateral, including CAD). All three already bundle CAD into the base code — there is no separate CAD add-on any more, a change dating to the 2017 CPT overhaul. The 2026 Medicare national average payment for 77065 is $124.54 globally ($37.93 professional component + $86.60 technical component), per published 2026 coding guides; actual amounts vary by Medicare Administrative Contractor and geography, and private payers typically run 15-20% above the Medicare base.

Software cost is the smallest line in this model, not the largest. AI-CAD platforms such as ScreenPoint Medical's Transpara are typically licensed per exam, at roughly £0.60 to £3.00 (about $0.76-$3.80) per study depending on volume and 2D-versus-3D mix, per NICE's technology assessment. Mature, well-run practices in this category report 15-28% net margins.

How AI-CAD changes the staffing math

The financial case for AI-CAD isn't really about the software fee at all - it's about reader throughput. Published research on AI-CAD-enhanced synthesized mammograms found radiologists could review digital breast tomosynthesis studies roughly 23.5% faster without a drop in interpretation accuracy. At the 6,000-study scale used in the worked example above, a 20%+ reduction in average read time is the difference between needing a second interpreting radiologist on staff and covering the volume with one radiologist plus a per-study overflow contract - a difference worth tens of thousands of dollars a year in fixed staffing cost, which dwarfs the software licence fee in either direction.

Worked example: what 6,000 exams a year actually generates

A single-site practice reading 6,000 CAD-enhanced screening and diagnostic studies annually, at a blended reimbursement of roughly $145 per study (Medicare's $124.54 base plus a typical 15-20% private-payer uplift), generates approximately $870,000 in annual imaging revenue. Against that, AI-CAD licensing at Transpara's published per-exam range adds only $4,560-$22,800 a year — under 3% of top-line revenue even at the high end. That ratio is why the constraint on this business is almost never software cost; it's radiologist availability, accreditation timing, and referral-channel development.

Worked example: a lean, phase-one launch

A smaller practice with a single suite reading 3,000 studies a year at the same $145 blended rate generates approximately $435,000 in year-one revenue. Against operating costs of roughly $310,000-$350,000 (technologist salary, radiologist read fees at typical per-study contract rates, facility lease, equipment maintenance, and AI-CAD licensing of $2,280-$11,400 a year at this lower volume), this scale generally lands close to breakeven in the first full year rather than generating meaningful profit — which is exactly why the phased-launch approach in the funding section above treats year one as a proof-of-volume phase, not a profit phase. Margin only becomes real once volume clears roughly 5,000-6,000 studies a year and fixed radiologist and facility costs are spread across more billable exams.

Self-pay and cash-price screening

Alongside insurance-billed volume, most independent centres also offer a cash-pay screening option for patients without coverage or who prefer a faster self-referral route without a physician order in states that permit it. Cash-pay screening mammograms are typically priced close to the blended insurance rate - commonly in the $150-$250 range in the US - rather than at a discount, since the marginal cost of an additional study (technologist time, AI-CAD licence fee, radiologist read fee) is largely fixed regardless of payer type. A plan should model this as a smaller, separate revenue line rather than folding it into the insurance-billed forecast, since payment timing (immediate) and volume predictability differ meaningfully from insurance receivables.

Licensing & Regulatory Requirements

Licensing for a CAD or AI-mammography practice spans facility accreditation, the coding/billing system, and — in the UK — device-level software regulation for the AI component itself. Getting the sequencing right matters more than any single requirement: most of the delays independent operators report come from ordering equipment before accreditation paperwork is filed, rather than from any single requirement being unusually slow on its own.

United States

• Mammography Quality Standards Act (MQSA) facility accreditation and certification, administered by the ACR — $5,000-$20,000 per modality, 60-120 days for initial accreditation
• CPT coding compliance for the bundled codes (77065/77066/77067) — CAD is no longer separately billable, a change that trips up plans built on older financial models
• State radiologic technologist licensure
• ACR-credentialed interpreting physicians and payer-network credentialing
• OSHA bloodborne pathogen and radiation-safety compliance

United Kingdom

• MHRA registration as Software/AI as a Medical Device (SaMD/AIaMD) for any proprietary algorithm, or verified UKCA/CE conformity evidence for a licensed third-party AI-CAD tool — typically £10,000-£40,000 in conformity-assessment cost depending on device risk class, 6-12 months
• Care Quality Commission (CQC) registration for the diagnostic/screening service, generally a 10-12 week process
• Alignment with the UK National Screening Committee's 2025 interim guidance on AI in breast screening if the practice intends to bid into NHS Breast Screening Programme (NHSBSP) contracts
• Professional indemnity insurance appropriate to diagnostic imaging
• Information governance and GDPR compliance for patient imaging data

International

• Australia: Therapeutic Goods Administration (TGA) registration of the AI/CAD tool as a software-based medical device, plus BreastScreen Australia accreditation for providers reading within the national screening programme
• Canada: Provincial diagnostic-facility licensing plus Health Canada medical device licensing for the AI-CAD software
• EU: CE marking under the Medical Device Regulation (MDR) for the software component, plus country-specific facility accreditation

The single biggest cross-border difference worth flagging early in any multi-jurisdiction plan is who regulates the algorithm itself. In the US, the FDA has historically treated CAD as sitting largely outside direct MQSA equipment regulation because it's applied after image acquisition rather than during it — the regulatory weight instead falls on facility accreditation and physician credentialing. In the UK, Australia, Canada and the EU, by contrast, the AI-CAD software is explicitly regulated as a standalone medical device (SaMD), which means the facility itself has to hold or verify device-level conformity evidence for whichever AI platform it runs, on top of the usual facility-level registration. A plan operating in more than one of these jurisdictions needs to budget for both regulatory tracks separately rather than assuming one accreditation covers both.

Common Mistakes First-Time Operators Make

Most of the plans that stall in underwriting fail for one of a small set of reasons, and nearly all of them are specific to how this category has changed since 2017 rather than generic startup missteps. The list below is drawn from the same gaps lenders and clinical reviewers flag most often in first-draft CAD/AI-mammography plans.

• Treating AI-CAD as a second reader instead of a triage tool. Clinical guidance and most payers still expect a qualified radiologist's sign-off on every study; a plan that implies AI replaces double-reading, rather than supporting it, will draw scrutiny from clinical reviewers and can also complicate malpractice-insurance underwriting.
• Underestimating the radiologist staffing gap. Roughly 4 in 5 practices report a shortage of breast-imaging radiologists. Recruitment and retention need to be modelled and budgeted for before equipment is even ordered - a signed interpreting-radiologist agreement, not just a hiring plan, is what actually satisfies a lender.
• Treating software integration as an afterthought. PACS and AI-CAD integration delay - not licensing cost - is the single most common reason a launch slips past its target date. Vendor integration timelines should be requested and cited in the operations plan, not assumed.
• Reusing a pre-2017 financial model. Older templates still assume CAD is separately billable under its own CPT add-on code; it has been bundled into 77065/77066/77067 since 2017, which overstates revenue by double-counting a line that no longer exists if not corrected.
• Sizing demand off stale screening guidelines. The 2024 USPSTF update lowered the recommended starting age from 50 to 40, meaningfully expanding the addressable population; plans still using the 50+ assumption undercount their own catchment area and can undersell the opportunity to a lender.
• Skipping the accreditation runway in the launch timeline. The 60-120 day MQSA/ACR accreditation window has to sit before, not alongside, the revenue-start date in the financial model - otherwise the cash-flow forecast will show income the business isn't yet legally able to bill for, which is one of the fastest ways to lose credibility with an underwriter.
• Ignoring the phased-capital option. Founders who ask for the full $620K upfront, rather than sequencing a smaller phase-one raise against proven volume, tend to face longer underwriting cycles and more collateral scrutiny than those who present a staged plan.

Sample Business Plan Preview

What's Inside the Template

Every Avvale business plan template includes these sections, pre-structured for a CAD/AI-mammography launch:

• Executive Summary — the practice at a glance, written to hold a lender's attention in the first sixty seconds
• Company Overview — legal structure, ownership, site selection, and founding rationale
• Industry Analysis — CAD/AI market data, vendor landscape, and the regulatory shifts reshaping demand
• Reimbursement & Revenue Model — CPT code mapping, payer mix, and per-exam economics
• Accreditation & Licensing Plan — MQSA/ACR timeline (US) or MHRA/CQC pathway (UK), sequenced against the launch schedule
• Marketing & Referral Strategy — referring-physician outreach, payer-network entry, and patient acquisition channels
• Operations Plan — staffing, scheduling, workflow between technologist, AI-CAD, and radiologist
• Management Team — founder and radiologist credentials, advisory relationships, key hires planned
• Vendor & Technology Selection — a short-list of AI-CAD platforms with selection criteria, so lenders see a considered choice rather than a single unexplained brand
• Risk Register — the accreditation-timeline, staffing-shortage, and reimbursement-policy risks specific to this category, each with a named mitigation

The optional Financial Forecast add-on (included in our $300/£250 and $1,000/£800 packages) provides a 5-year Excel model with income statement, cash flow, balance sheet, break-even analysis, and startup capital requirements sized to the equipment and accreditation timeline above. For a phased launch, the model separates phase-one and phase-two capital and revenue so lenders can see exactly when a second tranche becomes necessary and what utilisation level triggers it.

Frequently Asked Questions