The radiologist shortage is a workforce fact: imaging volume and the number of people who can sign a report are not growing at the same rate, and they are not in the same places. This page is that gap. It is not “AI will replace radiologists.” It is not a teleradiology job description. It is not a PYCAD staffing product.
If you meant what imaging AI actually does in a department → artificial intelligence in radiology (5644 already points here as the workforce URL). If you meant the remote reader → what is a teleradiologist. If you meant a named CT foundation model → Merlin. If you meant report-review as a benchmark → RADAR.
PYCAD builds custom web DICOM viewers and imaging models. A faster hang or a model that flags a candidate can change minutes per study. It does not hire a radiologist.
What “shortage” actually means
Two mismatches, usually at once:
- Headcount vs volume. More series, more slices per series, more after-hours. The training pipeline is residencies per year, not a hiring slider.
- Where the people are. High-income cities concentrate readers. A rural hospital and a low-income country can have a scanner and no one to read it. That is access, not a waiting list at one academic center.
RSNA News described the same squeeze in 2022: demand for imaging outrunning trainee output, plus burnout after COVID (Henderson, May 2022). That is the existence proof. It is not a headcount.
Numbers that have a name
| Claim | What it actually is | Treat it as |
|---|---|---|
| ~345,000 readers worldwide / ~45 per million (2023) | A Collective Minds tally (updated 2026; cites an analysis by Gassan Azem). Same piece: 794 readers in low-income countries / <2 per million | A company recap, not a WHO census. Order-of-magnitude only |
| “42,000 US radiologists short by 2033” | AAMC “other specialties” band (17,100–41,900) that includes radiology, pathology, psychiatry, and others. AAMC would not split radiology out | Do not quote as a radiology-only deficit. Dropped as a headline |
| US shortage persists through 2055 | Harvey L. Neiman HPI companion studies in JACR: radiologist supply +25.7% (2023–2055, no extra residency growth) vs imaging use +16.9% to +26.9% by modality. Shortage stays unless residencies grow or per-person use falls. Neiman HPI | The named US projection. Not “it gets 42,000 worse” |
| 69% of US practices understaffed (2023) | RBMA / ACR workforce survey, reported 2024 | A survey of members, not a census |
| WHO 2050: 22% of people over 60 | Standard WHO ageing figure. More old patients → more imaging. It does not by itself count radiologists | Demand pressure, not a headcount |
“Attrition up 50% since 2020” had no source on the old page. Dropped. Neiman HPI does model a higher post-COVID attrition rate; that is a scenario, not a 50% headline.
What actually closes a gap (and what does not)
- More trainees. Residencies are the supply dial. Everything else is a workaround.
- Teleradiology. Moves a reader to where the study is. It does not create a reader. The role URL is what is a teleradiologist.
- Assistive AI. A mark on a study the radiologist still signs. 5644 is that job. Cleared imaging AI is mostly SaMD assist, not autonomous. Merlin and RADAR are named research, not a staffing plan.
- A better viewer. Hanging, priors, a window that does not fight you. Minutes, not FTEs.
A vendor deck that says “our platform solves the shortage” is selling a workaround as a hire. PYCAD will not say that.
What this page is not
- Not 5644 restated with a shortage costume. That URL is what the model does; this one is why the inbox is full.
- Not a PYCAD staffing desk, a locums marketplace, or a “unified imaging platform” that replaces a department.
- Not
/portfolio. Named work is case studies. - Not a lighthouse metaphor, an HVAC outbound, or an unsourced burnout percentage. Dropped.
If the missing piece is a web viewer or a model that sits next to the worklist, that is the imaging piece. It is not a radiologist. Case studies.
Keep Reading
Related Articles
Explore the full DICOM HubAll services, tools, and guides on this topic — in one place.
Visit Hub →