How to read a radiology report

The single most useful habit in reading a scan report is to skip to the end. Every radiology report has a section called the Impression, and it is the radiologist's summary — the prioritized bottom line, written for the doctor to act on. RadiologyInfo, the patient-education site run by the radiology profession itself, says it plainly: the Impression "is the most important part of the radiology report for you and your doctor." It is where the radiologist tells you what, out of everything they looked at, actually matters.

Above the Impression sits the Findings section, and this is where most of the unnecessary fear comes from. Findings is a long, region-by-region catalog of everything the radiologist examined — the lungs, the liver, the kidneys, the bones, the soft tissue — and it dutifully notes the normal structures alongside any abnormal ones. On a completely reassuring scan, the Findings can still run for two or three dense paragraphs full of anatomical terms, because describing a normal liver still takes a sentence. Read top to bottom, that wall of text reads like a list of problems. It usually is not. It is a list of things checked.

So the move is simple: when a report opens, scroll to the Impression and read that first. If the Impression says something like "no acute abnormality" or "stable compared to prior," the long Findings section above it is almost certainly just the radiologist being thorough. If the Impression flags something specific, that is the line to bring to your doctor — and now you know which of the many sentences above it the radiologist thought was worth your attention.

One caution: reading the Impression first is for orientation, not for self-diagnosis. The Impression is written in shorthand and sometimes ends with a recommendation — a follow-up scan, a comparison with old images, a suggestion to correlate with lab work. Those are notes to your care team, not instructions for you to carry out alone. The point of reading it first is to find the one thing that matters quickly, so you can stop scanning the scary middle and start forming a question.

Once you know to read the Impression first, the rest of the report stops being a wall and becomes five labeled blocks. Whether the scan was a CT, an MRI, an X-ray, an ultrasound, or a PET study, and whoever the hospital is, the structure is the same one — the American College of Radiology's parameter for how findings are communicated lays out these components, and nearly every report follows them in order.

The first block is the clinical history, sometimes called the indication or reason for exam. It is one line saying why the scan was ordered — "52-year-old with abdominal pain," "follow-up of known lung nodule" — and it tells you the question the radiologist was trying to answer. The second is the comparison, which lists the prior studies this scan was read against. A line like "compared to CT from March 14" matters more than it looks: it means the radiologist is judging this scan against your history, not in isolation, which is the whole basis of follow-up imaging.

The third block is the technique — a few sentences on how the scan was done, including whether contrast was used. "With and without contrast" or a mention of gadolinium (the contrast agent used in MRI) is routine; contrast helps certain tissues and blood vessels stand out and does not, by itself, mean anything was found. The fourth is the Findings — the detailed observations from step one. And the fifth is the Impression, the summary you already know to read first.

That is the entire architecture. History, comparison, technique, findings, impression. When you can name the five parts, a report that looked like an undifferentiated block of jargon becomes a document you can navigate — you know where the question is, where the bottom line is, and that the long part in the middle is a catalog, not a verdict. Keeping every scan in one place, each report whole and dated, is what lets you find any of these sections again when the next appointment comes.

Most of the fear in a scan report comes from a handful of words that mean something specific to a radiologist and something much worse to everyone else. Learn this short glossary and a large share of the dread disappears. Start with the quietly reassuring ones. "Unremarkable," "within normal limits," and "no acute findings" all mean the radiologist did not see anything concerning in the area described — "unremarkable" is radiology for "normal," not for "we found nothing useful." "Stable," "unchanged," and "no interval change" mean the finding looks the same as it did on a prior scan, which is generally good news; as one radiology-practice paper put it, "no change or improvement is reassuring to the physician and patient."

Then there is the hedging language, which sounds like a warning but is usually just honesty about the limits of an image. "Cannot be excluded" or "cannot rule out" does not mean a condition is present — it means the scan cannot completely disprove a possibility, so the radiologist leaves it on the table for the doctor to consider. "Nonspecific" means the radiologist sees something real but its appearance could have several different causes, so they will not commit to one from the image alone. "Clinically correlate," or "clinical correlation recommended," is a request to your doctor to weigh the picture against your symptoms, exam, and lab results — it is asking for context, not raising an alarm. Radiologists hedge deliberately, because imaging is genuinely uncertain and overstating confidence would be the real error; the qualified language is the profession being careful, not evasive.

One phrase deserves its own paragraph because it frightens people more than almost any other: "incidental finding," sometimes called an incidentaloma. It means the scan turned up something unrelated to why it was ordered — a small cyst on a kidney seen on a scan for back pain, a tiny lung nodule found on a heart study. Incidental findings are extremely common and, in most cases, harmless. They feel like a bombshell precisely because no one was looking for them, but the great majority are benign and either need nothing or get a single follow-up to confirm they are not changing.

A few descriptive words also trip people up because they sound like diagnoses when they are not. A "lesion," a "nodule," and a "mass" are neutral terms for something the radiologist sees, not verdicts on what it is — a nodule is simply a small spot and a mass is a larger one, and a radiologist uses these words precisely because they do not presume cancer or anything else. On an MRI you may also see "T1" and "T2," which are just two different ways of taking the picture that make different tissues light up; a phrase like "T2 hyperintense" describes how something looked on one of those settings, not how dangerous it is. As with the rest of the glossary, the word names what was seen and leaves the meaning to the Impression and your doctor.

The thread running through all of these is that a radiology report is written in the careful, qualified language of a specialist documenting uncertainty — not in the plain language of a diagnosis. A word that reads as a verdict is usually a hedge, and a word that reads as dismissive ("unremarkable") is usually the best news on the page. When a phrase stops you, the right move is to look up what it means in this specific, narrow sense — or to ask the AI that has read the whole report to put it in plain English with the source line cited — rather than to let the worst reading stand unchallenged at midnight.

If you are reading scans during cancer care, the report carries a second vocabulary on top of the first, and it is the part almost no general guide explains. The most important thing to know is that in follow-up imaging, "stable" is usually good news. A line like "stable disease" or "no significant interval change" means the cancer is neither growing nor shrinking meaningfully since the last scan — which, during treatment, often means the treatment is holding the disease in check. It is not the dramatic win that "gone" would be, but for many people on long-term therapy, stable is exactly the goal.

"No evidence of disease," sometimes written NED, and "no evidence of metastatic disease" mean the scan found no detectable cancer. This is genuinely good news, and it is worth understanding precisely what it claims: no cancer is currently visible on imaging. It is not a guarantee that no microscopic cells remain anywhere, which is why oncologists tend to say "no evidence of disease" rather than "cured" — the scan can only speak to what it can see. The flip side is "progression" or "interval progression," which means the cancer has grown or spread since the prior scan. These are the words to bring straight to your oncologist; they are also the reason the comparison line from step two matters so much.

RECIST (Response Evaluation Criteria in Solid Tumors) is the standardized ruler that formalizes that comparison — the system oncologists use to judge whether treatment is working. A few target tumors are measured on each scan and compared against the baseline. Your oncologist applies the thresholds, not you, but knowing them demystifies the report: roughly, a 30 percent shrinkage in the summed measurements is a partial response, a 20 percent growth (and at least 5 millimeters) or any new lesion is progression, and everything in between is stable disease. This is why your scans are always "compared to prior," and why bringing your old images to a new facility is not a formality — without the baseline, the new scan cannot be scored.

Two more terms clear up a common confusion. A surveillance scan is a routine, scheduled scan after treatment, done to catch any return early while it is still small. A restaging scan is different — it is ordered when cancer is suspected to have come back or when the team needs to reassess how far it has spread, and it may lead to a new stage being assigned. Knowing which one you are reading changes how to hold it: a clean surveillance scan is the quiet, expected good news of a normal checkup; a restaging scan is answering a sharper question. Either way, keeping every scan and its priors together is what makes the next "compared to prior" possible.

Some scans come back with a category number attached, and a number on a medical report can be the most frightening thing of all when you do not know the scale. These scores exist to standardize how radiologists communicate suspicion, so that a "4" means roughly the same thing from one hospital to the next. The three you are most likely to meet are BI-RADS, Lung-RADS, and PI-RADS, and they all work the same way: a low number is reassuring, a high number means more concern, and the middle is a watch-and-recheck zone.

BI-RADS is the breast-imaging score, on a 0-to-6 scale. A 1 is a normal mammogram and a 2 is a benign finding; the one that causes the most needless worry is a 3, "probably benign," which actually carries a greater than 98 percent chance of being non-cancerous — it simply earns a short-interval follow-up to confirm it is not changing, rather than a biopsy. A 4 is "suspicious" and a 5 is "highly suggestive of cancer"; those are the numbers that lead to a biopsy. A 0 just means the radiologist needs more images before deciding.

Lung-RADS does the same job for low-dose CT lung-cancer screening: categories 1 and 2 are negative or benign with a return in a year, a 3 is probably benign with a six-month recheck, and the 4 categories are increasingly suspicious and prompt faster follow-up or biopsy. PI-RADS is the prostate-MRI version, scored 1 to 5 for the likelihood of a clinically significant cancer — a 1 or 2 usually means a biopsy is not needed, a 4 or 5 generally means it is. (Brain-tumor MRI uses its own scale, BT-RADS, on follow-up scans.)

The practical takeaway is the same across all of them: find your number, place it on the low-to-high scale, and remember that the lower and middle categories are designed to be conservative — they recheck rather than rush. A scoring number is a radiologist handing your doctor a shared language for what to do next. It is not a verdict, and the most common categories are the reassuring ones.

Once you can read the report, the question becomes what to do about it, and a flagged line sorts into three buckets. Knowing which one you are in is the difference between a sleepless night and a note to raise something next week. Read this as orientation, not medical advice — when in doubt, calling the nurse line is always reasonable, and your care team's read is the one that counts.

Call today, or use the after-hours line, when the Impression itself describes something acute or urgent — words like "acute," "new," a finding the report flags for prompt attention, or any result that frightens you in a scan ordered for a worrying symptom. Truly urgent findings are usually communicated to your doctor quickly by the radiologist, so if something is dangerous, a call to you is often already on its way. If you are not sure whether a line qualifies, that uncertainty is itself a good reason to call and ask.

Ask at the next visit covers the large middle: an incidental finding the report says to follow up in a few months, a "nonspecific" note, a scoring number in the low or middle range, a finding described as "likely benign." These are real and worth raising, but they are conversations, not emergencies. Write the question down — "the report mentioned a small nodule and said to recheck in six months, what does that mean for us" — and bring it. A specific question gets a specific answer.

Probably nothing is the most common bucket and the hardest to sit with, and here the data genuinely helps. Incidental findings are not rare events — depending on the population scanned they are common, with one widely cited estimate putting them on around 45 percent of chest CT scans, and the vast majority are benign. The trouble is that chasing them has a real downside: in a 2019 survey of nearly a thousand physicians, almost all had seen an incidental finding set off a "cascade" of follow-up tests, and most reported it caused their patients psychological harm. The lesson is not to ignore your report — it is to resist the late-night spiral. A single incidental finding with no urgent language, on a scan that answered its actual question, is usually exactly what the word implies: incidental. This is where it helps most to have something that has read the whole report in context, can explain a flagged line in plain English with the source cited, and does not diagnose — it orients, and leaves the judgment with your clinician.

You so often read the report before anyone calls because of one specific change. Since the 21st Century Cures Act's information-blocking rules took effect in April 2021, results, radiology reports included, are released to your patient portal as soon as they are finalized, frequently before the ordering doctor has reviewed them. Patients overwhelmingly want it that way: in a large 2023 study, more than 95 percent preferred immediate access even before their clinician had looked, though the same study found the obvious cost, that abnormal results came with a markedly higher likelihood of worry. A separate 2025 study found that more than 40 percent of sensitive results are now seen by patients before their doctors. Seeing your scan first is normal, not a sign anything was missed. It just means the literacy this guide builds has to do some work in the gap before the call.

The most useful thing that literacy unlocks is the trend. A single scan is a snapshot; the meaning usually lives in the comparison. A lung nodule measured at the same size across three scans is a different story from one that grew, and the radiologist can only tell you which by reading this scan against the last. That is why the most valuable thing you can do as the keeper of the records is to keep every scan and its report together, dated, so the next "compared to prior" is always possible. A digital medical binder that reads each report as it arrives turns a folder of PDFs into a legible history — the same study tracked over time, the priors ready when a new facility asks for them.

Bringing the right question is the other half. An appointment is short, and the difference between a useful one and a wasted one is often a single well-formed question prepared in advance. Reading the Impression ahead of time — sorting it into the three buckets, noting the one line that moved — lets you walk in with "the report says the nodule is stable compared to March, should we keep the same follow-up schedule" instead of handing over a printout and hoping. If you waited anxiously for the scan in the first place, the companion guide on how long test results take explains why the wait varies and what a quiet stretch does and does not mean.

A last word on the failsafe. Any tool that summarizes a scan — including a good one — should show its work. A plain-English summary you cannot trace back to the radiologist's actual words is a guess dressed up as an answer, and a guess is the last thing a family in a hard stretch needs more of. Insist on the citation, check any summary against the real Impression, and treat the radiologist's report as the truth and the summary as the convenience. That habit, more than any glossary, is what makes reading your own scans safe to do. And when the result is a lab panel rather than a scan, the same calm approach applies — the companion guide on reading your lab results walks through reference ranges the same way this one walks through the Impression.