Incidentalomas, How Radiologists Decide What Is Worth Chasing

The short answer

An incidentaloma is a finding no one was looking for, a spot the scan turned up while imaging was ordered for something else entirely. Radiologists do not chase all of them, and good radiology holds that they should not. The decision is a probability judgment: how likely is this particular finding, in this particular patient, to be something that would change care, weighed against the cost and harm of pursuing it. Over the past fifteen years the American College of Radiology (ACR) has worked to convert that judgment from instinct into published, organ-by-organ algorithms, so that what gets chased rests on evidence rather than reflex.

Why incidental findings became a management problem

Cross-sectional imaging keeps getting better at showing things, including things unrelated to the reason for the study. A 2018 umbrella review in The BMJ by O'Sullivan and colleagues, pooling twenty systematic reviews, found that the frequency of incidental findings ranges from under 5 percent in some settings, such as whole-body PET, to more than a third in others, such as chest CT and cardiac MRI. Just as important, the chance that an incidentaloma is actually malignant swings widely by organ, from under 5 percent for adrenal and brain findings to roughly 42 percent for breast findings. That single fact is the reason a blanket rule cannot work.

The variability creates a two-sided risk. Miss the rare early cancer and a patient loses a real chance. Chase every shadow and you generate biopsies, follow-up scans, radiation, cost, and anxiety for findings that were never going to matter. The ACR frames its incidental-findings work as balancing the benefit of catching important disease while patients are still asymptomatic against the harm of over-testing findings that carry very low risk.

Turning judgment into algorithms

The foundational document is the 2010 ACR Incidental Findings Committee white paper on managing incidental findings on abdominal CT, which set out practical pathways for the kidneys, liver, adrenal glands, and pancreas. A second committee extended the approach to adnexal, vascular, splenic, nodal, gallbladder, and biliary findings, and later organ-specific papers updated the renal and adrenal recommendations. The series is not static. The ACR continues to add pathways, with guidance on incidentally discovered pineal cysts appearing as recently as 2025, and it maintains the whole collection as a living reference clinicians can pull up at the workstation. Parallel efforts outside the ACR do the same job for other organs, most prominently the Fleischner Society guidelines for pulmonary nodules.

Each pathway reads like a decision tree. It asks a short series of questions about the finding, such as whether it is cystic or solid, how big it is, whether it carries features that are reassuring, and whether the patient has a history that raises the stakes, then routes to one of a few endpoints: ignore it, describe it and stop, follow it at an interval, or work it up now.

The Bayesian core

Underneath the flowcharts sits Bayesian reasoning, even where the papers never use the word. Every recommendation starts from a pretest probability, the baseline likelihood that a finding of this type is dangerous, and then updates that estimate using imaging features and patient context. Size thresholds are the clearest example. The Fleischner Society raised its minimum follow-up size for solid pulmonary nodules to 6 mm precisely because screening data showed the cancer risk below that size sits well under 1 percent, so routine follow-up cannot help enough to justify its harms. The number is chosen to keep the probability of missing something meaningful acceptably low.

Imaging features shift the estimate further. A renal lesion containing macroscopic fat points toward a benign angiomyolipoma. A simple cyst with water density and a thin, smooth wall needs no chasing at all. An adrenal nodule that measures 10 Hounsfield units or less on an unenhanced scan is very likely a lipid-rich adenoma. Each of these is a likelihood ratio in disguise, a feature that moves the estimate toward benign strongly enough to close the workup.

What tips a finding into worth chasing

Findings earn a workup when the features push probability the other way. Solid rather than cystic, enhancement after contrast, a thick or irregular wall, size above the organ-specific threshold, and above all growth compared with a prior study all raise concern. Patient context weighs as heavily as pixels. The same small lung nodule that warrants nothing in a lifelong nonsmoker may warrant a follow-up CT in a heavy smoker, which is why the Fleischner recommendations are stratified by risk in the first place.

This is also why the cheapest and most powerful test in radiology is often the oldest scan. A nodule that has been stable for years is reassuring in a way no single image can be, and comparison with priors resolves a large share of incidentalomas without any new procedure.

Where the logic strains

The algorithms are only as good as the data behind them, and for some organs that evidence remains thin, which is why several pathways offer ranges and discretion rather than fixed commands. There is also the deeper problem of overdiagnosis. A pathway that finds and confirms an indolent lesion has still generated worry, cost, and sometimes an invasive procedure for a condition that would never have surfaced on its own. The ACR guidance is written to shrink that harm, not to erase the uncomfortable truth that more looking finds more of everything, useful and useless alike.

Understanding how these decisions are made can make an incidental line on your own report less alarming, but it is educational information, not medical advice, and the workup for any specific finding belongs with the clinician who ordered the scan.