Radiology’s history reveals a recurring pattern: Innovation arrives first, understanding follows. From X-rays to artificial intelligence (AI), each breakthrough generated new questions that only accumulated evidence could help answer. This has formed a clear pattern of innovation followed by understanding.
That is why access to historical literature remains essential for today's researchers. The Radiological Society of North America (RSNA) has built that evidence base since Radiology® began regular publication in 1923, documenting changes in imaging technology, terminology, and clinical understanding. This record matters because some of radiology's current research questions reach further back than the language researchers now use to ask them.
A century of imaging advances and the evidence they created
Each major advance in medical imaging created new clinical possibilities and a new set of research questions. As new technologies entered practice, researchers had to establish their limitations and refine how they should be used:
- 1950s: Pulse-echo methods became the basis of modern diagnostic ultrasound, bringing radiation-free imaging to medicine.
- 1970s: Computed tomography (CT) introduced cross-sectional imaging, leading researchers to establish faster acquisition methods, image quality standards, and safe radiation dose.
- 1970s–1980s: Magnetic resonance imaging (MRI) introduced a non-ionizing approach to soft-tissue imaging, followed by research into hardware, reconstruction methods, and clinical applications.
- 1990s–2000s: Radiology moved from film archives to searchable digital systems, changing how the field stores, retrieves, and distributes medical images.
Each transition changed the questions researchers asked, the methods they used, and some of the vocabulary used to describe findings. As those conventions shifted, earlier evidence became harder to retrieve through contemporary search terms.
Why old radiological research can still answer new questions
Older studies can trace how a diagnostic approach, method, or body of evidence developed. This may require literature published before current terminology and research conventions existed.
For instance, a research question can outlast the terminology originally used to describe it. Citation searching allows researchers to follow references backward to earlier studies and forward into earlier work, tracing relationships a keyword search may miss.
Earlier studies can also show why current protocols developed as they did, by letting researchers look into how a diagnostic approach or evidence changed over time. Looking across a longer record can help researchers establish how a field reached its current understanding.
For example, a team studying lung cancer imaging protocols may follow this process:
- Reviewing recent findings through current subscriptions
- Identifying landmark studies that established earlier diagnostic approaches
- Tracing how those studies were cited and challenged over time
- Reviewing how diagnostic protocols shifted as new evidence entered the scene
The useful time horizon for a research question depends on the evidence it requires, and historical coverage becomes necessary once that evidence extends beyond current publications.
What role does historical research play in the AI era?
Interpreting new AI findings within a broader evidence base shapes how clinically meaningful evidence is evaluated in the age of AI. Published research, imaging datasets, and AI training data serve different purposes, and treating them as interchangeable is where AI-assisted research may end up inaccurate and unsuitable for the research question.
- Published research documents questions, methods, findings, limitations, and interpretations over time.
- Imaging datasets contain observations used for research and analysis.
- Training data provide examples from which an AI model learns patterns.
Historical radiology literature can show which methods researchers used, the comparison points they established, and the limitations they identified. That evidence helps researchers judge what a newer AI study contributes to an existing body of work.
AI also changes how researchers work with published evidence itself. Large language models are increasingly used to accelerate evidence synthesis, a task that would otherwise require researchers to process large volumes of literature manually. However, processing volume isn’t the same as judging quality. Human evaluation is required to establish the quality and applicability of an AI-assisted study.
RSNA's Checklist for Artificial Intelligence in Medical Imaging (CLAIM) provides a reporting framework to improve reproducibility in medical imaging AI research, covering research objectives, dataset handling, model evaluation, and limitations.
How RSNA's published record connects past and present
RSNA’s journal portfolio combines historical depth with coverage across current research and education, including:
- Radiology®: Current research across medical imaging
- RadioGraphics®: Peer-reviewed educational material for radiologists, trainees, and physicists
- Radiology: Artificial Intelligence®: AI, machine learning, and data science applied to imaging
- Radiology: Cardiothoracic Imaging®: Advances in cardiac, vascular, and thoracic imaging research
- Radiology: Imaging Cancer®: Imaging research related to cancer screening, diagnosis, and treatment planning
Longer-running publications provide historical depth while newer specialty journals document questions emerging across contemporary radiology.
The RSNA Complete Collection on Ovid brings these titles together to provide searchable access to decades of radiology research and education through a one-time purchase. Its breadth and depth allow researchers to move across specialties or back through years of evidence as their questions evolve.
Why a full radiology record matters for your institution
Research institutions increasingly support multidisciplinary teams working across oncology, cardiology, data science, and AI. These teams often need access to different eras of literature at the same time, which makes historical depth an institutional capability, rather than simply a library asset.
New imaging methods create possibilities. Research then establishes where they work, where they fall short, and how they should be applied — a pattern that has held for over a century and will likely continue as new technologies emerge. Preserving access to historical radiology literature helps ensure new discoveries build on prior evidence instead of starting from scratch.
Explore the RSNA Complete Radiology Collection on Ovid and request a free trial to see how current research and historical evidence work together across specialty imaging, education, and AI.