The 3-phase adrenal washout CT protocol uses unenhanced, 60-second venous, and 15-minute delayed phases to separate benign adenomas from adrenocortical carcinoma, pheochromocytoma, and metastatic disease using validated HU and washout thresholds.

3-Phase Adrenal Washout CT Protocol: 7 Critical Steps for Accurate Mass Characterisation

1. Introduction: why adrenal washout CT still matters

The 3-phase adrenal washout CT protocol remains one of the most heavily relied-upon problem-solving tools in abdominal imaging. Adrenal nodules are identified incidentally on a substantial proportion of routine abdominal CT examinations performed for unrelated indications, and the overwhelming majority of these are benign, non-functioning cortical adenomas. The clinical challenge is not detecting the adrenal mass — it is confidently separating that common benign adenoma from the rare but consequential adrenocortical carcinoma, metastasis, or pheochromocytoma without subjecting every patient to biopsy or unnecessary surveillance.

This protocol exploits a simple biological fact: benign cortical adenomas are typically rich in intracytoplasmic lipid, which lowers their baseline attenuation, and they take up and then rapidly release iodinated contrast media faster than most other adrenal masses. By acquiring an unenhanced phase, a 60-second venous phase, and a 15-minute delayed phase, the radiology team can calculate quantitative washout percentages that have been validated across decades of literature and remain embedded in current European and American guidelines.

Despite its long track record, washout CT has come under renewed scrutiny. Recent re-analyses of the foundational literature have questioned whether the 10 HU and washout thresholds were ever properly validated in true incidentaloma populations, and 2023 European Society of Endocrinology (ESE) guidance has shifted toward earlier reliance on hormonal screening.^[1,2] This article presents the protocol as it is most widely practised today, while flagging where the evidence base is evolving — because radiographers, radiologists, and referring physicians all need to understand not just how the numbers are generated, but how much weight they should be given.

What follows is a complete operational reference: adrenal anatomy and HU values, the full seven-step scanning technique across 16- to 320-slice platforms, the contrast injection protocol, radiation dose benchmarks, the ten pathologies most likely to appear on an adrenal washout study, and — critically — a three-tier pitfall framework covering the distinct failure modes of radiographers, radiologists, and the non-radiology physicians who order and act on these reports.

The scale of the problem: why this protocol exists

Adrenal incidentalomas are detected on a meaningful fraction of all abdominal CT scans performed for entirely unrelated reasons, with reported prevalence rising with patient age and scan resolution. The vast majority — typically well over 80% in unselected incidentaloma cohorts — turn out to be non-functioning, benign cortical adenomas of no clinical consequence. Yet within that population sits a small minority of adrenocortical carcinomas, functioning tumours secreting cortisol, aldosterone, or catecholamines, and metastatic deposits from a primary malignancy elsewhere in the body.^[4,9]

This statistical reality is precisely why a quantitative, reproducible imaging test matters so much. Sending every incidentally discovered adrenal nodule to biopsy or surgical excision would expose a very large number of patients with entirely benign disease to unnecessary procedural risk, cost, and anxiety. Conversely, dismissing every adrenal nodule as “probably an adenoma” without any quantitative threshold would inevitably miss a proportion of clinically important disease. The 3-phase washout protocol sits precisely at this decision point, translating a qualitative visual impression into a number that can be tracked, audited, and compared against a published threshold.

It is worth being explicit, however, that this is an area of active scientific debate. Recent large reanalyses — including a 2025 publication provocatively titled to question whether washout CT’s era is ending — have argued that several of the foundational washout studies were conducted in selected surgical or oncology populations rather than true unselected incidentaloma cohorts, and that real-world diagnostic performance in general radiology practice may be lower than the original thresholds imply.^[1,7] This article presents the protocol as it is currently and widely practised, while being transparent about where the evidence is shifting, so that radiographers, radiologists, and clinicians can apply appropriate clinical judgement rather than mechanical rule-following.

There is also a straightforward health-economic argument for getting this protocol right. A single avoidable indeterminate result — caused by a truncated delayed phase, an inconsistent ROI, or an overly hedged report — does not simply disappear; it typically generates a follow-up MRI, a repeat CT at a later interval, an endocrinology referral, or in some cases a biopsy, each carrying its own cost, scheduling burden, and patient anxiety. Multiplied across the volume of adrenal incidentalomas a busy department encounters annually, the cumulative cost of avoidable indeterminate results is a meaningful departmental quality metric in its own right, distinct from the headline diagnostic accuracy of the protocol itself.

2. Adrenal anatomy and Hounsfield unit reference values

The adrenal glands sit in the retroperitoneal fat superomedial to each kidney, draped over the diaphragmatic crura. The right adrenal gland is typically pyramidal or triangular and lies posterior to the inferior vena cava, lateral to the right diaphragmatic crus, and superior to the right kidney’s upper pole. The left adrenal gland is more crescentic or arrowhead-shaped, lying medial to the upper pole of the left kidney, posterior to the pancreatic tail and splenic vessels, and lateral to the aorta. Each normal gland measures roughly 2–6 cm in length with limbs (the medial and lateral crura) typically less than 10 mm in thickness; any limb thickening beyond this raises suspicion for hyperplasia or infiltrative disease.

Functional anatomy: cortex and medulla

Each gland is composed of an outer cortex and inner medulla with distinct embryological origins, hormone products, and — critically for this protocol — distinct lipid content:

• Zona glomerulosa (outermost cortex) — produces aldosterone; the source of primary aldosteronism when adenomatous.
• Zona fasciculata (middle cortex, the bulk of the gland) — produces cortisol; the most lipid-rich layer and the principal reason cortical adenomas are often low-attenuation on unenhanced CT.
• Zona reticularis (innermost cortex) — produces androgens.
• Medulla — neural crest-derived chromaffin tissue producing catecholamines; the origin of pheochromocytoma.

Vascular supply and venous drainage

The adrenal gland has a unique triple arterial supply: the superior adrenal artery from the inferior phrenic artery, the middle adrenal artery arising directly from the aorta, and the inferior adrenal artery from the renal artery. Venous drainage is asymmetric and clinically important for adrenal vein sampling: the right adrenal vein drains directly and very short into the posterolateral inferior vena cava, while the left adrenal vein drains inferiorly into the left renal vein, typically joined by the left inferior phrenic vein. This asymmetry explains why right-sided adrenal vein catheterisation is technically more demanding during aldosteronoma localisation studies.

This same triple-supply, single-dominant-vein anatomy is directly relevant to laparoscopic and robotic adrenalectomy planning, where the surgical team relies on preoperative CT to anticipate vessel position and anatomical variants before the operative approach is chosen. A radiology report for a surgical candidate that simply states “adrenal mass present” without describing its relationship to the adjacent renal vessels, IVC, or splenic vessels provides materially less value to the operating surgeon than one that explicitly addresses these relationships, particularly for larger masses where vascular displacement or encasement changes the surgical approach.

Embryological origin and why it matters on imaging

The cortex and medulla arise from entirely different embryological tissues, and this distinction is not merely academic — it explains the imaging behaviour of every pathology discussed later in this article. The cortex develops from mesodermal coelomic epithelium adjacent to the developing gonad, which is why ectopic adrenocortical tissue can occasionally be found along the gonadal descent pathway. The medulla is neuroectodermal in origin, derived from migrating neural crest cells that also give rise to the sympathetic paraganglia — the same embryological lineage shared by extra-adrenal paragangliomas, which is why pheochromocytoma and paraganglioma are considered part of a single biological spectrum despite differing in anatomical location.

Lymphatic drainage

Adrenal lymphatic drainage follows the venous pathways closely: on the right, lymphatics drain toward the lateral aortic and retrocaval nodal groups; on the left, drainage follows the left renal vein toward the left para-aortic and renal hilar nodes. This matters directly for adrenocortical carcinoma staging, where nodal involvement along these specific chains should be actively sought on the same study, rather than relying on a generic “no significant lymphadenopathy” statement.

Relevant clinical anatomy: adrenal vein sampling and percutaneous biopsy approach

When a functioning aldosteronoma is suspected, adrenal vein sampling (AVS) is frequently required to lateralise the source of excess hormone production before surgery, since CT morphology alone is an unreliable predictor of which gland — or which nodule within a gland — is functionally responsible. The marked asymmetry in venous anatomy described above makes the right adrenal vein technically the more difficult target to catheterise selectively, owing to its short length and oblique, often variable, entry point into the posterolateral IVC; this is one of the most common reasons AVS procedures are reported as technically unsuccessful or non-lateralising.

For percutaneous biopsy of an indeterminate adrenal mass — reserved for cases where imaging and biochemical workup remain genuinely inconclusive — the approach is typically posterior or posterolateral, threading between the lung base, diaphragm, and kidney while avoiding the spleen on the left and the liver and IVC on the right. Biopsy is rarely required when the 3-phase washout protocol yields a clear benign result, which is precisely the cost-saving and risk-reducing rationale for performing the protocol correctly in the first place.

Normal variants and anatomical pitfalls

Several normal anatomical variants are routinely mistaken for pathology on adrenal CT, and recognising them prevents unnecessary downstream workup. A prominent, broad-based “Mercedes-Benz” or inverted-Y configuration of the limbs is a normal variant in many patients and should not by itself be read as nodular enlargement. Adjacent structures are also frequently misread as adrenal pathology: a tortuous splenic artery or vein looping near the left gland, a prominent gastric diverticulum adjacent to the left adrenal bed, or an accessory spleen sitting in the expected location of the left gland can all mimic a left adrenal nodule on a single axial slice. Multiplanar reformatting — reviewing coronal and sagittal reconstructions rather than relying on the axial plane alone — resolves the great majority of these pseudo-lesion pitfalls before they reach the report.

Right-sided pseudo-nodules are less common but occur where a prominent diaphragmatic crus or an adjacent hepatic lesion abutting the gland creates partial-volume averaging that suggests a discrete adrenal mass. In every case, the practical safeguard is the same: confirm that the suspected lesion is genuinely intrinsic to the gland on at least two imaging planes before committing it to a formal washout calculation, since an ROI placed partly on adjacent fat or vessel will produce a meaningless and potentially misleading attenuation value.

3. Scanning technique: the 7-step protocol

The 3-phase adrenal washout protocol is unforgiving of timing errors. Every step below is built around protecting the integrity of the three attenuation measurements that the entire diagnosis depends on.

1. 1
   Patient preparation and positioning. Position the patient supine, arms raised above the head to eliminate streak artifact across the upper abdomen. Confirm fasting status if local policy requires it, screen renal function (eGFR) and prior contrast reaction history, and place a large-bore (18–20 gauge) antecubital intravenous line capable of sustaining a 3.0 mL/s injection. Explain the full duration of the examination to the patient before beginning — including the 15-minute interval between the second and third phases — since patients who understand they will be lying still, intermittently, for over twenty minutes in total are considerably less likely to move or request early removal from the table than those who were told only “this will be quick.”
2. 2
   Unenhanced acquisition. Acquire a true non-contrast scan through the adrenal beds at 120 kVp, 150–250 mA with automatic exposure control, 0.5 second rotation time, and pitch 1.0. This unenhanced phase is non-negotiable — without it, only the relative (not absolute) washout percentage can be calculated, and the 10 HU lipid-rich threshold cannot be applied at all.
3. 3
   Region-of-interest planning on the unenhanced phase. Place a region of interest (ROI) of at least 1 cm² (avoiding partial-volume averaging at the lesion margin) on the unenhanced image and record the mean HU. This value becomes the fixed reference point for every subsequent washout calculation, so it must be placed identically on each subsequent phase.
4. 4
   Contrast injection. Power-inject 95 mL of low-osmolar non-ionic iodinated contrast at 3.0 mL/s, immediately followed by a 100 mL saline chaser at the same flow rate to maximise bolus compactness and reduce residual contrast dead-space in the tubing and central veins.
5. 5
   Venous phase acquisition at 60 seconds. Acquire the second phase at a fixed 60-second delay from the start of injection (not a bolus-tracking trigger, since the adrenal glands are not the primary target of arterial enhancement). Re-measure the identical ROI and record the venous HU.
6. 6
   Delayed phase acquisition at 15 minutes. Acquire the third and final phase at exactly 15 minutes post-injection. This interval is the single most common point of protocol failure in busy departments — flag the patient, set a timer, and do not allow table turnover pressure to compress this window.
7. 7
   Washout calculation and documentation. Calculate the absolute percentage washout (APW) as [(venous HU − delayed HU) ÷ (venous HU − unenhanced HU)] × 100, and the relative percentage washout (RPW) as [(venous HU − delayed HU) ÷ venous HU] × 100. Document all three raw HU values alongside the calculated percentages in the report — not just the final classification — so any subsequent reviewer can audit the calculation.

Dual-energy and photon-counting CT adaptations

Dual-energy CT (DECT) has matured into a genuinely useful adjunct for adrenal characterisation rather than a research curiosity. By generating virtual unenhanced (VUE) images directly from the contrast-enhanced venous phase, DECT can in principle estimate the unenhanced attenuation without a separate non-contrast acquisition — directly reducing dose. However, validation studies report important caveats.

The underlying physics matters for how these caveats manifest in practice. Dual-source and rapid kVp-switching systems acquire two genuinely separate energy spectra nearly simultaneously, while dual-layer detector systems separate low- and high-energy signal at the detector level from a single x-ray source and acquisition — each approach has slightly different noise characteristics and material decomposition accuracy, which is precisely why virtual unenhanced thresholds validated on one platform have not transferred cleanly to another without local recalibration. A department adopting VUE-based adrenal characterisation for the first time should not assume the published 10 HU lipid-rich threshold applies unmodified to their specific vendor and software version; published validation work has shown the adjusted threshold can sit closer to 20–22 HU on some dual-layer platforms once a systematic positive offset is accounted for.^[18]

Deep learning reconstruction (DLR)

Deep learning reconstruction algorithms — distinct from AI lesion-classification tools discussed later in this article — denoise the raw projection or image data itself, allowing tube current or kVp reduction while preserving diagnostic image quality. For adrenal protocols specifically, DLR is most valuable on the unenhanced phase, where noise has historically forced technologists to choose between a higher dose and a noisier ROI measurement. Departments adopting DLR on 64-slice-and-above platforms have generally been able to reduce unenhanced-phase mA by 20–35% without measurably degrading ROI reproducibility, though formal adrenal-specific validation data remains limited and vendor-dependent; confirm performance with your own phantom or retrospective audit before relying on it clinically.

Equipment calibration and reproducibility across scanners

Because this protocol depends on absolute HU values being comparable across three separate acquisitions, and ideally across separate examinations over time for surveillance purposes, scanner calibration is not a peripheral physics concern but a direct determinant of diagnostic reliability. Routine CT quality assurance programmes should specifically verify HU accuracy and stability for water and standard reference phantom materials at the kVp used in this protocol, and any scanner showing drift outside locally defined tolerance should be flagged before it is used for adrenal washout studies, since a systematic HU offset of even a few units can shift a genuinely borderline lesion across the benign threshold in either direction.

For departments running adrenal washout protocols across multiple scanners — common in larger hospital networks — establishing that HU measurements are reasonably consistent between platforms, and ideally cross-calibrated against a shared phantom, materially increases confidence when a follow-up examination for the same patient is performed on a different machine than the original study.

4. Contrast media protocol

Because the entire diagnostic logic of this protocol depends on a contrast time-attenuation curve, the injection parameters are not interchangeable with a generic abdominal CT protocol. Consistency between the venous and delayed measurements requires a bolus delivered the same way, every time.

Pre-procedure renal function screening follows standard institutional contrast safety protocols; patients with significantly reduced eGFR may require modified hydration or, where clinically appropriate, an unenhanced-only limited protocol with deferral of washout calculation. Patients with a history of severe contrast reaction require premedication per local policy or consideration of an alternative non-contrast pathway (chemical-shift MRI) for adenoma characterisation where lipid-rich features are suspected.

Monitoring during the 15-minute interval

The mandatory 15-minute gap between the venous and delayed phases is not idle time. It is the window in which the injection site should be visually checked for delayed extravasation, the patient should be monitored for any delayed contrast reaction symptoms, and venous access should be confirmed as still patent in case the delayed phase reveals a need for an additional targeted acquisition. Departments that treat this interval purely as a scheduling inconvenience — rushing the patient out of the room and back in — both increase the risk of the primary scanning pitfall (a truncated delay) and lose a built-in safety monitoring opportunity.

Venous access should be re-confirmed, not assumed, immediately before the delayed acquisition; a line that has tissued or a patient who has needed to reposition can silently undermine the comparability of the delayed-phase measurement if a second, smaller “rescue” bolus is inadvertently given rather than recognising the line has failed.

When CT washout is not the right tool: MRI and special patient groups

Chemical-shift MRI offers an alternative, non-ionising route to characterising lipid-rich adenomas, exploiting the same intracytoplasmic lipid signature that drives the low unenhanced HU on CT, but detecting it through signal drop-out between in-phase and opposed-phase sequences rather than attenuation. This is a particularly useful alternative pathway for patients in whom repeated CT radiation exposure is a concern — young patients with an incidentaloma requiring interval surveillance, or patients who have already accumulated substantial cumulative dose from prior oncological staging studies. MRI does not, however, reliably replace CT washout for lipid-poor adenomas, where the contrast kinetics captured by the delayed CT phase have no direct MRI equivalent with comparable validation.

Pregnant patients with a suspected adrenal mass requiring urgent characterisation — most commonly in the context of suspected pheochromocytoma presenting with hypertensive crisis — should generally be directed toward MRI without gadolinium-based contrast in the first instance, reserving CT washout for situations where MRI is unavailable, contraindicated, or non-diagnostic, and always in direct discussion with the obstetric and endocrine teams regarding the urgency-versus-radiation trade-off. Paediatric adrenal masses differ sufficiently in their differential diagnosis — neuroblastoma being a key consideration rarely relevant in adults — that this adult-oriented washout protocol should not be applied unmodified to children without paediatric radiology input on both technique and interpretation.

5. Radiation dose benchmarks and reduction

A 3-phase protocol inherently delivers more radiation than a single-phase abdominal CT, since the adrenal beds are imaged three separate times. This makes dose optimisation at each phase — particularly the unenhanced phase, which contributes diagnostic value disproportionate to its dose if performed correctly — a genuine clinical priority rather than a box-ticking exercise.

These figures are indicative benchmarks consistent with the framework set out in ICRP Publication 135 and the European Society of Radiology literature review on CT diagnostic reference levels^[13,14]; departments should establish and audit their own local DRLs against national survey data rather than treating any single published figure as a universal target, since scanner generation, reconstruction algorithm, and patient body habitus all materially shift the achievable dose for equivalent image quality.

Understanding the dose metrics in this table

CTDIvol describes the dose delivered to a standard reference phantom and is the metric most directly comparable across scanners and institutions, but it does not represent dose to an individual patient. Size-specific dose estimate (SSDE) corrects CTDIvol for the patient’s actual cross-sectional dimensions, and is the more clinically meaningful figure when comparing dose between a slender and an obese patient undergoing the identical nominal protocol. DLP integrates dose over the full scan length and is the figure used to estimate effective dose via region-specific conversion factors — useful for population-level risk communication, but, like all effective dose estimates, an approximation rather than a patient-specific measurement.

For a 3-phase protocol specifically, the cumulative DLP across all three acquisitions is the single most useful number to track at a departmental level, since it captures the full radiation burden of the complete examination rather than any individual phase in isolation. Departments performing high volumes of adrenal washout studies should periodically audit their own cumulative DLP distribution against the benchmark ranges above, flagging outlier cases for review rather than waiting for an external accreditation cycle to surface a systematic dose creep.

Five dose reduction strategies

1. Limit the anatomical coverage of each phase. The venous and delayed phases do not need to cover the entire abdomen and pelvis if the unenhanced phase has already localised the lesion — a focused adrenal-bed acquisition on the later phases meaningfully reduces total DLP.
2. Apply automatic exposure control (AEC) consistently across all three phases. Manually fixing mA “for consistency” across phases often means over-radiating the smaller-coverage delayed acquisition; AEC should be allowed to modulate per phase based on attenuation.
3. Adopt iterative or deep learning reconstruction on the unenhanced phase specifically. Because the unenhanced HU measurement is the most noise-sensitive of the three (low-contrast, low-attenuation tissue), modern reconstruction algorithms allow meaningful mA reduction here without compromising the 10 HU threshold’s reliability.
4. Consider omitting the unenhanced phase when prior imaging exists. If a genuinely unenhanced CT of the same lesion exists in the patient’s history within a clinically reasonable timeframe, some departments avoid re-acquiring it — though this requires confident, auditable access to the prior study and should follow a documented local policy, not ad hoc clinician discretion.
5. Use dual-energy virtual unenhanced reconstruction where validated locally. As detailed in the scanning technique section, DECT-derived virtual unenhanced images can in principle replace the physical unenhanced acquisition entirely, removing one full phase of radiation — but only after local threshold validation, since reported sensitivity varies by vendor and detector technology.^[17,18]

Quality assurance and departmental audit

Beyond per-patient dose tracking, departments performing this protocol at meaningful volume benefit from periodic structured audit: sampling a quarter’s worth of adrenal washout studies and confirming, independently of the original report, that the unenhanced phase truly preceded contrast, that the venous-to-delayed interval was genuinely 15 minutes (not just labelled as such in the protocol name), and that cumulative DLP figures sit within the locally agreed benchmark range. This kind of retrospective audit is one of the few reliable ways to catch a systematic workflow drift — for example, a scheduling change that has quietly compressed the delayed-phase interval across an entire department — before it accumulates into a meaningful body of mis-classified studies.

6. Top 10 adrenal pathologies on washout CT

Adrenal incidentalomas span a spectrum from entirely benign and clinically irrelevant through to rare, aggressive malignancy. The cards below summarise the ten conditions most likely to be encountered on a 3-phase adrenal washout study, and — critically — how each one interacts with (or defeats) the washout calculation itself.

Building a differential diagnosis from the washout calculation

None of these ten entities is reliably separated from the others by a single isolated data point. A practical reporting approach treats the unenhanced HU, the washout percentages, the lesion size, the margin characteristics, and the clinical history as a single integrated picture rather than a checklist applied in isolation. A 3 cm, smoothly marginated, homogeneous nodule with unenhanced attenuation of 8 HU requires no further discussion of washout at all — it is diagnostic of a lipid-rich adenoma on the unenhanced phase alone. A 5 cm heterogeneous mass with central necrosis and an absolute washout of 58% — technically just below the 60% benign threshold — should not be reported as “probably benign, just under threshold,” because the morphology alone in a mass of this size already raises adrenocortical carcinoma as a leading concern regardless of the washout number.^[3,12] This is the central interpretive skill the radiologist pitfalls section below addresses in detail.

A third illustrative scenario worth holding in mind: a 1.8 cm right adrenal nodule, stable in size over two prior CT examinations spanning several years, with an unenhanced attenuation of 24 HU and an absolute washout of 52%. Taken in isolation at a single time point, this falls into genuinely indeterminate territory by the conventional thresholds. Taken together with the documented multi-year size stability, most experienced reviewers would reasonably favour a benign, slowly evolving lesion over an aggressive process — illustrating why access to, and active comparison with, prior imaging is frequently as diagnostically valuable as the washout calculation performed at any single visit.

7. Pitfalls for radiographers

The primary scanning pitfall in this protocol is failing to wait the full 15 minutes for the delayed phase. Cutting the delay short prevents accurate washout calculations, leaving an otherwise classifiable mass indeterminate — and triggering exactly the follow-up imaging or biopsy this protocol exists to avoid.

This pitfall is particularly insidious because it rarely produces an obviously wrong-looking image. A delayed phase acquired at 10 or 11 minutes instead of 15 still looks like a perfectly reasonable delayed-phase scan to the eye — the error only becomes apparent at the calculation stage, when the washout percentage comes out lower than it should, and by then the patient has often left the department. Departments under throughput pressure are especially vulnerable to this failure mode, because a 15-minute mandatory gap per adrenal patient represents genuine lost table time that scheduling systems do not always account for. The mitigation is therefore as much about departmental workflow design — building the 15-minute interval explicitly into the appointment slot length — as it is about individual technologist vigilance.

8. Pitfalls for radiologists

The primary interpretation pitfall is that hypervascular metastases or pheochromocytomas can occasionally demonstrate rapid washout profiles, mimicking benign adenomas. Washout CT is a probabilistic tool, not a categorical one, and treating a favourable washout percentage as definitive in the wrong clinical context is a recognised failure mode.

The underlying biological reason is that washout reflects contrast kinetics in general — the rate at which a vascular bed takes up and then releases iodinated contrast — and is not a direct measure of lipid content or histological benignity. Lipid-rich tissue happens to wash out quickly, which is why the threshold works well for the majority of adenomas. But other tissue types with high microvascular permeability and rapid interstitial clearance can wash out quickly for entirely different physiological reasons, producing a false sense of reassurance. Reported series suggest this overlap occurs in a meaningful minority — historically cited at up to around a quarter — of pathologically confirmed pheochromocytomas, which is precisely why biochemical correlation cannot be skipped on the basis of a reassuring CT alone.^[10,11]

None of these pitfalls is an argument against using washout CT — they are an argument for using it as one input among several, reported with appropriate transparency about its own limitations. A radiologist who reports “absolute washout 71%, favouring adenoma, though biochemical correlation is recommended given [specific clinical detail]” has done meaningfully more for patient safety than one who reports a bare “benign adenoma” — the additional sentence costs almost nothing in reporting time and materially changes how the next clinician in the pathway interprets and acts on the result.

9. Pitfalls for non-radiology physicians

Referring and treating physicians who order or act on adrenal washout CT reports face a different category of risk: not misreading the images, but misapplying or over-trusting the conclusion without the radiological nuance behind it.

This is, in many respects, the highest-stakes pitfall category in the entire protocol, because it sits closest to the point of clinical decision-making. A radiographer’s timing error or a radiologist’s interpretive nuance can usually be caught and corrected within the imaging pathway itself — a second technologist notices the timer was wrong, or a peer review flags an overly confident report. A referring physician’s misapplication of a washout result, by contrast, often only surfaces much later, when a patient who should have had biochemical screening presents instead with an undiagnosed phaeochromocytoma crisis, or a hormonally active adenoma is found incidentally years after it could have been treated.

The common thread across all five of these clinical pitfalls is a mismatch between what an imaging report can and cannot establish. A washout CT report answers a specific, narrow question — does this mass behave like a benign cortical adenoma on contrast kinetics? — and was never designed to answer the broader questions of hormonal function, malignancy risk in the context of a specific cancer history, or safety for an invasive procedure. Referring physicians who keep this narrow scope explicitly in mind, rather than treating “the CT looked fine” as a comprehensive clearance, are far less likely to fall into any of the five pitfalls above.

10. Pitfall comparison summary

What this comparison makes visible is that the three pitfall tiers are sequential, not parallel — an error at the scanning stage propagates forward and constrains what the radiologist can possibly conclude, and an overly confident or ambiguous radiologist’s report propagates forward again into how the referring physician acts. A single robust quality safeguard at any one tier (a reliable delayed-phase timer, a habit of stating raw HU values in every report, a departmental policy of treating “indeterminate” as an action item) meaningfully reduces the chance of a downstream failure two steps removed from where the safeguard was applied.

11. AI and automation in adrenal CT

Artificial intelligence applications in adrenal imaging have moved well beyond proof-of-concept. A 2025 systematic review of CT-based radiomics in adrenal lesion characterisation found pooled diagnostic performance with a mean area under the curve of approximately 0.88 across eligible studies, with the strongest-performing models reaching an AUC of 0.99 for identifying aldosterone-producing adenomas.^[31] These are research-stage and early clinical-validation results rather than universally deployed clinical tools, and departments should treat published AUC figures as evidence of promise rather than guaranteed real-world performance on their own patient population and scanner fleet.

Several distinct AI application categories are emerging for this protocol specifically:

• Automated lesion segmentation. Deep learning segmentation models reduce inter-observer variability in ROI placement — directly addressing one of the radiographer and radiologist pitfalls described above — by generating a reproducible, algorithm-defined region rather than a manually drawn one.^[30]
• Texture and radiomic feature classification. Beyond simple mean HU, radiomic models extract heterogeneity, edge, and texture features from unenhanced and contrast-enhanced images to separate lipid-poor adenomas from non-adenomas, including in cases where conventional washout values are borderline.^[20,21]
• Deep learning reconstruction (DLR). As discussed in the scanning technique section, DLR denoises image data to support dose reduction, particularly valuable on the noise-sensitive unenhanced phase.
• Automated volumetric adrenal gland measurement. Fully automated tools for measuring adrenal gland volume support more objective hyperplasia classification than manual limb-thickness callipers, which carries known measurement-technique variability.^[23]

Practical integration considerations

Even where a validated AI tool exists, integrating it into an adrenal washout workflow raises practical questions that are easy to overlook during procurement. Does the tool require the unenhanced phase, or is it designed to work from virtual unenhanced reconstructions alone? Does it report a binary classification, a continuous probability score, or a set of extracted radiomic features that still require human synthesis? Can it flag discordance between its own output and the conventional washout calculation for radiologist review, rather than silently overriding it? These questions matter more for adoption success than the headline AUC figure quoted in a vendor’s marketing material.

A further consideration specific to this protocol is that several of the most promising radiomic and deep learning classifiers were trained and validated on data from a single triphasic acquisition pattern, often using a specific scanner vendor’s reconstruction kernel. A tool validated on one institution’s 64-slice, 120 kVp dataset may perform differently on a 320-slice volumetric acquisition or a dual-energy-derived virtual unenhanced image from a different vendor. Local validation against a department’s own historical cases — even a modest retrospective audit of 50–100 known outcomes — remains good practice before any AI output is trusted to influence clinical decision-making, regardless of the regulatory status of the underlying product.

Multidisciplinary coordination and documentation

Adrenal incidentaloma management increasingly sits at the intersection of radiology, endocrinology, and — when malignancy is confirmed or strongly suspected — surgical and oncological teams. Departments performing high volumes of adrenal washout studies benefit from a defined pathway specifying exactly which findings trigger automatic endocrinology referral (any biochemical abnormality, any mass >4 cm, any indeterminate washout result), rather than leaving this judgement to vary case-by-case based on which radiologist or physician happens to be involved. A documented multidisciplinary adrenal tumour board, even one that meets relatively infrequently, materially improves consistency in management of the genuinely ambiguous cases that this protocol, by design, cannot resolve through imaging alone.^[27]

From a documentation and medico-legal standpoint, the single most protective habit described throughout this article is the same one repeated across the pitfall tables: record the raw numbers, not just the conclusion. A report that states “absolute washout 64%, consistent with adenoma” allows any future reviewer — including a different radiologist reading a follow-up scan, or a clinician reassessing the case in light of new symptoms — to independently verify and, if necessary, challenge the original classification. A report that states only “adenoma” forecloses that possibility entirely, and in the rare case where the original classification proves wrong, removes the audit trail that would explain why. This single habit, more than any individual threshold or technology discussed in this article, is the most reliably protective practice available to every member of the imaging and clinical team.

12. Further reading

1. Multi-Phase Liver CT Protocol: 7 Critical Steps for HCC Detection
2. Dual-Phase Pancreatic CT Protocol: 7 Critical Steps for Mass Detection
3. Routine Abdomen Pelvis CT Protocol: 7 Proven Scanning Steps
4. 2026 Worldwide Guidelines for Safe Contrast Media Administration: eGFR Thresholds and Society Recommendations
5. The Price We Pay for Bubbles in CT and MRI: Understanding Venous Air Embolism in Contrast-Enhanced Imaging

13. Conclusion

The 3-phase adrenal washout CT protocol distils a complex diagnostic question — is this incidentally discovered adrenal mass dangerous? — into a structured, reproducible sequence of three precisely timed acquisitions. Its diagnostic power rests entirely on disciplined execution: a true unenhanced phase, a 60-second venous phase, and an uncompromised 15-minute delayed phase, each measured with a consistently placed region of interest.

Across the ten pathologies most commonly encountered — from the reassuringly common lipid-rich adenoma through to the rare but lethal adrenocortical carcinoma — the protocol’s value depends on recognising its limits as much as its strengths. Pheochromocytoma can defeat the washout calculation outright; hypervascular metastases can mimic benign kinetics; and the underlying threshold literature itself is under active re-evaluation by major endocrine societies.

The three-tier pitfall framework presented here — scanning execution for radiographers, interpretive nuance for radiologists, and clinical application for referring physicians — reflects the reality that no single specialty can deliver this protocol’s value alone. Consistent timing at the scanner, careful interpretive language in the report, and appropriate biochemical correlation by the treating clinician together determine whether a patient with an adrenal incidentaloma is reassured appropriately, investigated appropriately, or — in the rare but consequential cases — referred for definitive management without unnecessary delay.

As dual-energy and photon-counting platforms mature, and as radiomic and deep learning classifiers move from research validation toward genuine clinical deployment, the operational shape of this protocol will likely continue to evolve — potentially toward fewer acquired phases and more derived measurements. What will not change is the underlying clinical discipline this article has emphasised throughout: precise timing, transparent documentation of the underlying numbers, and a willingness to treat a borderline result as exactly that, rather than forcing it into a falsely reassuring binary outcome.

For radiology departments building or refining this protocol, the practical priorities are clear: protect the 15-minute delayed-phase interval in scheduling and workflow design, standardise ROI placement methodology across staff, document raw HU values alongside any calculated percentage, and maintain an active, two-way relationship with referring endocrinology and oncology teams so that indeterminate results are tracked through to resolution rather than left as an open loop in the patient’s record.

14. References