How Inconsistent Are Imaging Procedure Names Across Health Systems?
Very. In a Medicom study of 124 US healthcare organizations, six routine exams appeared under 1,258 distinct descriptions, and only 9.1% of organization pairs shared even one exact description for the same exam. More than 1 in 4 sided exams left out laterality, and contrast phase showed up in 10 different notations.
Radiology has spent two decades building shared vocabularies for naming imaging procedures. LOINC and the RSNA RadLex Playbook define a common language, DICOM constrains the metadata that travels with every study, and nearly every health system keeps a curated internal procedure catalog. What nobody had measured is how well that work holds up when a patient's imaging crosses from one organization to another. Medicom moves imaging across that boundary every day, so we measured it.
How Did Medicom Measure Procedure Naming?
The study used two views of real network traffic, with no patient identifiers:
- Prior Search Data: Over seven days, users at 30 organizations ran 22,143 prior imaging searches through Medicom Smart Search against 113 partner institutions, 124 organizations in all. Those searches produced 404,748 procedure description instances and 17,262 distinct strings.
- DICOM Object Data: 1,345,763 DICOM objects received through Medicom's SMART on FHIR applications, a largely separate customer population, showed how the Modality field behaves in the wild.
- Network Mix: The 124 organizations include 57 community, regional, and rural hospitals, 22 integrated delivery networks, 17 outpatient imaging providers, 12 academic health systems, eight oncology and specialty groups, six orthopedic groups, and two federal facilities across more than 15 states.
How Many Names Does One Imaging Exam Have?
A two-view chest X-ray, one of the most common exams in medicine, appeared under 145 distinct spellings across 78 organizations. The most common spelling, "X-RAY EXAM CHEST 2 VIEWS," covered just 26.8% of instances. The top three covered 61.5%, eight covered 80%, and full coverage took all 145.
Every routine exam the study examined looked the same:
- Screening Mammography: 392 distinct names across 70 organizations, an average of 5.6 per organization that performs it.
- CT Abdomen and Pelvis With Contrast: 234 names across 69 organizations.
- MRI Lumbar Spine Without Contrast: 204 names across 62 organizations.
- CT Head Without Contrast: 148 names across 70 organizations.
- Chest X-Ray, Two Views: 145 names across 78 organizations.
- Pelvic Ultrasound: 135 names across 47 organizations.
That's 1,258 distinct strings for six routine procedures in one week of traffic.
How Often Do Two Hospitals Use the Same Name for the Same Exam?
Rarely. Among 13,151 pairs of organizations holding descriptions for the same exam, only 9.1% shared even one identical string. Agreement ranged from 3.0% for screening mammography to 15.2% for lumbar spine MRI. Pick any two institutions that perform the same routine exam, and more than nine times in 10 they won't share a single spelling.
That means exact text matching between any two organizations fails by default. Misses are the normal case.
Which Naming Differences Matter Most Clinically?
The variation reaches details that change how a study is read:
- Laterality: Of 60,256 descriptions naming a sided body part such as a wrist, knee, or shoulder, 17,013, or 28.2%, carried no side at all. Those that did state a side split across RT/LT, RIGHT/LEFT, and BILAT/BILATERAL styles.
- Contrast Phase: Contrast showed up in 10 notations across 96,545 instances. WO, W/O, WITHOUT, and W O all mean without contrast. W WO, WITH AND WITHOUT, W/WO, W AND WO, WWO, and W&WO all mean with and without.
- Modality: The DICOM Modality field is supposed to draw from a short, defined list. The study found 121 distinct values, including lowercase variants, vendor strings, SUPPLIES, and UNKNOWN. Plain radiography alone arrived under three codes: CR, DX, and DR.
Truly invalid Modality values were rare, under 1% of objects in every source channel. The problem is the variety. If the most tightly governed field in the standard drifts this far, free text descriptions never stood a chance.
Do Hospitals Name Exams Consistently Inside Their Own Walls?
Not always. The study found 962 cases where a single organization wrote the same procedure more than one way, such as one academic health system using both "MRI CERVICAL SPINE WO CONTRAST" and "MRI SPINE CERVICAL W/O CONTRAST."
Simple cleanup doesn't fix it either. After folding case, punctuation, spacing, and common abbreviations, only 16.4% of distinct strings collapsed into another. The other 83.6% reflects real differences in local vocabulary that no formatting script can reconcile.
Internal catalogs still look tidy, because agreement inside one organization is high. The mess only shows up when studies cross organizational lines, which is why it's easy to miss and hard for any one IT team to govern. For more on why this happens, see Why Don't Hospitals Describe the Same Imaging Exam the Same Way?
What Does Inconsistent Procedure Naming Cost Patients, Providers, and Health Systems?
Every one of these strings eventually lands in front of a person or system that has to decide what it means.
- Patients: When a prior goes unfound, the scan gets ordered again, adding radiation, cost, and delay. The patients most affected are the ones whose care crosses boundaries: trauma transfers, oncology referrals, second opinions, and anyone whose imaging lives in more than one place.
- Providers: A radiologist reading a screening mammogram needs the right priors on screen. Hanging protocols key on the description to arrange them. When the vocabulary shifts, the comparison study is missing or out of order, and the read takes longer.
- Health Systems: Duplicate order checks, prior retrieval, and AI models trained on description strings all break quietly at the network boundary. Manual mapping files and compendiums were never built to keep up with 392 names for one exam.
How Should Health Systems Standardize Imaging Procedure Names?
The paper recommends five practices:
- Map at the Boundary: Standards such as the LOINC/RSNA Radiology Playbook deliver the most value where studies cross organizational lines. Exchange infrastructure should attach a mapped standard code to every study it moves while keeping the local description for provenance.
- Make Laterality and Contrast Structured: Both have homes in DICOM and order entry. Requiring them as discrete, validated fields at ordering, and not only at the modality worklist, closes the largest clinical gaps. Health systems first need to define which procedures require a side. CMS could reinforce this through billing rules.
- Measure Vocabulary Drift as a Quality Metric: Track distinct descriptions per procedure the same way you track report turnaround and radiation dose, both internally and against what arrives from exchange partners.
- Treat Relevancy as Infrastructure: Prior retrieval, hanging protocols, duplicate detection, and AI cohort selection should all match on clinical meaning. Teams buying enterprise imaging platforms should test retrieval across vocabularies directly.
- Preserve Source Fidelity: Normalize by annotating instead of overwriting. Receiving systems need both the local name and the shared translation so the mapping stays auditable and can improve over time.
How Does Medicom Handle Mismatched Exam Names?
Medicom Smart Search computes relevance on clinical meaning. In production, it returns a mean of 17.3 candidate priors per search and judges a mean of 3.6 relevant, a 79% reduction in what clinicians have to sort through. A search launched from one organization's "BI MAMMO SCREEN W TOMO" surfaces a partner's "BI MAMMOGRAM DIAGNOSTIC DIGITAL TOMOSYNTHESIS BILATERAL," even though the two strings share no exact match.
Read the full research report, or book a demo to see how Smart Search finds the right priors across vocabularies.
Frequently Asked Questions
How many names can one imaging exam have across hospitals?
In one week of Medicom network traffic, screening mammography appeared under 392 distinct names and a two-view chest X-ray under 145. Six routine exams added up to 1,258 distinct descriptions.
How often do two hospitals use the same description for the same exam?
Only 9.1% of organization pairs shared even one exact description for the same exam. Screening mammography was lowest at 3.0%, and lumbar spine MRI was highest at 15.2%.
How often is laterality missing from imaging procedure descriptions?
Of 60,256 descriptions naming a sided body part, 28.2% didn't state a side. Breast imaging was excluded from that count.
Can a cleanup script fix inconsistent procedure names?
Mostly no. After normalizing case, punctuation, spacing, and common abbreviations, only 16.4% of distinct strings collapsed together. The remaining 83.6% reflects real differences in local vocabulary.
What's the best way to standardize imaging procedure names across a network?
Map local names to a standard such as the LOINC/RSNA Radiology Playbook where studies cross organizations, keep the original description alongside it, make laterality and contrast structured fields, and match on clinical meaning instead of exact text.
More Answers

Answers
How Can a Health System Stop Relying on Imaging CDs?
Replace disc intake with electronic exchange that runs on top of the EHR, RIS, and PACS you already use. Medicom Connect links facilities over encrypted peer-to-peer connections, delivers outside studies into the EHR, and lets patients upload priors before appointments. Onsite Women's Health grew image transfer volume 13x after moving from eight exchange platforms to one.

Answers
How Does Faster Access to Prior Imaging Change ED Outcomes?
In a JAMIA study, every hour cut from the time it took to access outside patient information was associated with a 52.9-minute shorter ED stay, fewer CT, MRI, and X-ray orders, a 2.4-point drop in admissions, and $1,187 lower charges per visit. Prior imaging is where that access is slowest.

Answers
How Do You Get Outside Imaging Into the EHR Without a Manual Order?
Outside imaging can't attach to the EHR without a matching order and accession number, so it often sits unseen in the archive. Medicom's Auto Order Generation reads the study's DICOM metadata and sends an HL7 order or result automatically, so outside studies show up in the chart just like exams performed on site.
Your Plan for Enterprise Imaging Interoperability
Talk to one of our experts to plan your interoperability strategy, explore workflows, or get a personalized demo. We’re here to help.



