Reporting Clinically Actionable Imaging Features of Intracerebral Hemorrhage: A Survey of Neuroradiologist Practice

Abstract

Background and Purpose:

Recent advances in spontaneous intracerebral hemorrhage (ICH) management increasingly expand the role of neuroimaging beyond diagnosis to potentially guide time-sensitive therapies and targeted clinical trial enrollment. We surveyed neuroradiologists to characterize current ICH imaging and reporting practices and identify opportunities for quality improvement.

Materials and Methods:

An anonymous 52-item electronic survey was distributed through the American Journal of Neuroradiology Survey Corner and REDCap between April and May 2026. Survey domains included imaging acquisition, reporting practices, hematoma volume quantification, assessment of hematoma expansion predictors, ICH localization, etiologic assessment, and follow-up imaging. Responses were summarized using descriptive statistics.

Results:

A total of 267 neuroradiologists completed the survey. Nearly all respondents (97.3%) reported 24/7 emergency head CT interpretation coverage. However, 41.5% did not routinely quantify hematoma volume, only 45% reported routine CTA acquisition for patients with ICH, and 30% rarely or only occasionally recommended CTA when it was not performed. While 91.3% routinely reported a positive CTA spot-sign, only 43.0% almost always assessed non-contrast CT predictors of hematoma expansion. Confidence in hemorrhage localization when crossing anatomical boundaries and etiologic classification was modest, and 45.8% routinely used the Boston Criteria version 2.0 for MRI-based assessment of cerebral amyloid angiopathy.

Conclusions:

Contemporary neuroradiology practice demonstrates several readily addressable gaps in ICH imaging and reporting, including hematoma volume quantification, routine vascular imaging, standardized assessment of hematoma expansion imaging markers, and etiologic assessment. Standardizing these practices can improved image guided risk-stratification and facilitate implementation of emerging time-sensitive ICH therapies.

Introduction:

Spontaneous, non-traumatic intracerebral hemorrhage (ICH) accounts for 10–15% of strokes in the United States and 29% of strokes globally.^{1, 2} However, ICH is associated with a 30–40% mortality rate and contributes to nearly half of the global stroke-related disability burden.^{2, 3} Historically, ICH management has focused on supportive care and blood pressure control. However, this paradigm has shifted since 2024 with clinical trials reporting benefits of minimally invasive hematoma evacuation for lobar ICH (30–80 mL),⁴ combined decompressive craniectomy and medical treatment in severe (30–100mL) deep ICH,⁵ intensive care bundles incorporating blood pressure reduction and management of hyperglycemia, fever, and anticoagulation within 6 hours of onset,⁶ as well as ultra-early prehospital blood pressure lowering.⁷ Moreover, a tertiary analysis of a recombinant factor VIIa clinical trial suggested that hemostatic therapy may benefit ICH patients with a CTA spot-sign.⁸ Together, these advances underscore the need for a “Code ICH” workflow incorporating rapid diagnosis and identification of patients eligible for time-sensitive therapies.⁹

Neuroimaging plays a central role in acute ICH care as outlined in the 2022 American Heart Association/American Stroke Association (AHA/ASA) guidelines for ICH management³ and the 2024 AHA/ASA ICH Performance and Quality Measures.¹⁰ These guidelines strongly recommend (class 1) rapid head CT or MRI for ICH diagnosis. CTA, MRA, or catheter angiography is also recommended in select patients to exclude an underlying macrovascular lesion among those with lobar ICH who are <70 years old or with deep or posterior fossa ICH who are <45 years old or 45–70 years old without a history of hypertension. They also state that repeat head CT within 24 hours or more frequent serial imaging may be considered (Class 2b) in patients with depressed consciousness or neurological deterioration. For (non-urgent) diagnostic assessment, MRI is recommended to evaluate for cerebral amyloid angiopathy (CAA), which is associated with an estimated annual ICH recurrence risk of 7.4% compared with 1.1% for non-CAA hemorrhage and may influence decisions regarding antithrombotic therapy.¹¹ The guidelines also include a Class 2b recommendation to obtain CTA within the first few hours after presentation and evaluate for the spot-sign and that non-contrast CT scans be assessed for imaging markers associated with hematoma expansion to help identify at-risk patients.³ Finally, hematoma volume measurement, most commonly using the ABC/2 method, is recommended for clinical decision-making – specifically for surgical evacuation of ≥15 mL cerebellar ICH.^{3, 10}

Despite emerging time-sensitive therapies and evolving ICH management guidelines, contemporary neuroradiology practice for ICH imaging and reporting has not been systematically characterized. Beyond early detection of ICH, imaging assessment of hematoma volume, location, expansion risk, and underlying etiology directly informs acute treatment decisions. We therefore conducted an anonymous electronic survey of the neuroradiology community in United States and Canada, distributed through the AJNR Survey Corner and REDCap during April and May 2026, to characterize current ICH imaging and reporting practices. We report the findings in the context of the 2022 AHA/ASA ICH Guidelines and the 2024 Performance and Quality Measures,^{3, 10} highlighting practice gaps and their implications for guideline adherence and quality improvement.

Methods:

Survey design and administration:

An anonymous, cross-sectional electronic survey was developed to evaluate current neuroradiology practices for the diagnosis, interpretation, and reporting of spontaneous ICH. The 52-item questionnaire is provided in the supplemental materials, and included multiple-choice, multiple-select, and options for open-ended responses. The questionnaire was investigator-developed based on imaging recommendations in the 2022 AHA/ASA Guidelines for the Management of Spontaneous ICH and the 2024 AHA/ASA ICH Performance and Quality Measures.^{3, 10} Survey domains included respondent demographics and practice characteristics, imaging acquisition protocols, core reporting elements, hematoma volume quantification, assessment of radiologic predictors of hematoma expansion, anatomical localization of mixed-location hematomas, etiologic assessment of ICH, recommendations for follow-up imaging, and structured reporting practices. The primary imaging question was framed as evaluation of “suspected acute ICH”; subsequent questions regarding reporting practices, recommendations for CTA after non-contrast CT, hematoma-expansion markers, and follow-up imaging referred to patients with ICH.” Recommendations for follow-up imaging were queried broadly for “acute spontaneous ICH” and were not stratified by hemorrhage pattern or suspected etiology. The survey was administered using REDCap, and disseminated through the AJNR Survey Corner and via a publicly accessible REDCap link between April and May 2026. The AJNR Survey Corner link was distributed to the ASNR membership in monthly newsletter, and the survey was additionally sent by direct email invitation to 3,749 ASNR members. Participation was voluntary, anonymous, and uncompensated. Respondents were permitted to skip individual questions; therefore, denominators vary across survey items according to the number of valid responses. As a non-human subject study, no institutional review board approval was required.

Statistical analysis:

Survey responses were summarized using descriptive statistics. Categorical variables are presented as counts and percentages calculated from the number of valid (non-missing) responses for each survey item. Free-text responses were reviewed and incorporated into existing response categories when appropriate or summarized as new categories when they represented distinct themes. Because the primary objective of the study was to describe current practice patterns, no formal hypothesis testing or inferential statistical analyses were performed. In addition, respondents were de-identified, precluding identification of individuals from the same institution or differentiation of individual preferences from institution-level practices.

Results:

Respondent and practice characteristics:

Of 270 individuals who started the survey, 267 completed it (99% completion among starters). Unique Survey Corner link views were not available. Using the direct-email distribution as the denominator, the approximate completion response rate was 7.1% (267/3,749); because the Survey Corner link was publicly accessible and ASNR membership is not restricted to radiologists, the overall response rate specifically among the neuroradiology community could not be determined. As completion of individual questions was optional, some items had missing responses, and the denominator differs between analyses. As summarized in Table 1, most respondents were neuroradiologists (245, 91.4%), had ≥11 years of practice experience, worked at academic settings (153, 57.5%) and Comprehensive Stroke Centers (203, 77.2%), with ≤20 unique ICH case load per month (163/258, 63.2%). Among the 243 respondents who reported their practice location, 74 (30.5%) practiced in the Southern United States, 65 (26.8%) in the Northeast, 52 (21.4%) in the West, 41 (16.9%) in the Midwest, and 11 (4.5%) in Canada (Details in supplemental Table 1).

Table 1.

Respondent and practice characteristics.

Characteristic	Category	Number (%)
Primary specialty (n=267)	Neuroradiology	244 (91.4%)
	Diagnostic Radiology	14 (5.2%)
	Interventional Neuroradiology	8 (3.0%)
	Radiology resident	1 (0.4%)
Years in practice (n=265)	0–5	64 (24.2%)
	6–10	45 (17.0%)
	11–20	80 (30.2%)
	>20	76 (28.7%)
Practice setting (n=266)	Academic	153 (57.5%)
	Private	51 (19.2%)
	Hybrid/Academic-affiliated	42 (15.8%)
	Tele-radiology	10 (3.8%)
	Government/VA	8 (3.0%)
	Other (please specify)	2 (0.8%)
Stroke center designation (n=263)	Comprehensive stroke center	203 (77.2%)
	Primary stroke center	15 (5.7%)
	Thrombectomy-Capable Stroke Center	13 (4.9%)
	Acute Stroke Ready Hospital	12 (4.6%)
	Don’t know	20 (7.6%)
Monthly spontaneous ICH volume (n=258)	< 5	38 (14.7%)
	5–10	61 (23.6%)
	11–20	64 (24.8%)
	21–30	27 (10.5%)
	31–40	10 (3.9%)
	>40	58 (22.5%)

Percentages are of valid responses for each item listed in the left column.

Imaging acquisition and reporting:

Non-contrast CT alone (typically monoenergetic) was the predominant first-line imaging modality for suspected acute ICH (61.6%), whereas 37.6% of respondents routinely performed simultaneous CTA – with or without CT perfusion (Table 2). Nearly all respondents (97.3%) reported 24/7 emergency neuroradiology coverage for head CT interpretation (Table 2). Dual-energy (spectral) CT, either alone or combined with CTA, was reported by 13 respondents (4.9%). Nearly half of respondents (118/265, 44.5%) reported that CTA is routinely performed for all patients with ICH at their institution. Among the remainder, recommendations for CTA following non-contrast CT varied, with 30.2% recommending CTA only occasionally or rarely (Table 2). Regarding follow-up imaging recommendations in radiology reports (multiple responses permitted, Figure 1), the most common approach was no routine follow-up unless requested by the clinical team (41%), followed by MRI to evaluate for an underlying lesion (36%) and CTA/MRA (33%). Routine follow-up CT within 6 or 12 hours of symptom onset was recommended by only 24–25% of respondents (Figure 1). Most respondents used free-text reporting (230/266, 86.5%), with only 18 (6.8%) using institution-wide and another 18 (6.8%) using personal templates for structured reporting. Explicit recommendations for neurosurgical consultation were uncommon, with 59% reporting that they included such recommendations only rarely (Table 2).

Table 2.

Acute ICH imaging and report practices.

Characteristic	Response	Number (%)
Primary imaging modality for suspected acute ICH (n=263)	Non-contrast CT (conventional mono-energetic)	151 (57.0%)
	Dual-energy (or spectral) CT	12 (4.6%)
	Simultaneous non-contrast CT (mono- or dual-energy) and CTA	85 (32.3%)
	Simultaneous CT, CTA and CTP	14 (5.3%)
	Routine MRI with and without contrast	1 (0.4%)
24/7 radiology coverage for emergency head CT (n=264)	In-house 24/7	136 (51.5%)
	Hybrid	112 (42.4%)
	Teleradiology-only	9 (3.4%)
	No 24/7 in-house or teleradiology coverage	7 (2.7%)
Recommending CTA, when ICH patient only had non-contrast head CT (n=265)	Rarely (0–25%)	50 (18.9%)
	Occasionally (26–50%)	30 (11.3%)
	Frequently (51–75%)	26 (9.8%)
	Almost always (76–100%)	41 (15.5%)
	At our center, patients always get CTA	118 (44.5%)
Including neurosurgical consult recommendations in radiology report (n=266)	Rarely (0–25%)	157 (59.0%)
	Occasionally (26–50%)	57 (21.4%)
	Frequently (51–75%)	24 (9.0%)
	Almost always (76–100%)	28 (10.5%)
Hematoma volume quantification (n=265)	ABC/2 method (or three-dimension reporting)	108 (40.7%)
	Reporting a single or two dimensions	2 (0.8%)
	Automated software measurement	14 (5.3%)
	Visual estimate	31 (11.7%)
	Not routinely quantified	110 (41.5%)
Peri-hematomal edema measurement or assessment (n=263)	Report visual estimate	158 (60.1%)
	Only comment when radiologically significant	71 (27.0%)
	Comment when worsened over serial studies	23 (8.7%)
	Not routinely quantified or addressed in the report	11 (4.2%)
Intraventricular hemorrhage and/or hydrocephalus (n=266)	Rarely (0–25%)	1 (0.4%)
	Occasionally (26–50%)	1 (0.4%)
	Frequently (51–75%)	10 (3.8%)
	Almost always (76–100%)	254 (95.5%)

Percentages are of valid responses for each item listed in the left column.

Figure 1.

Follow-up imaging recommended for acute spontaneous ICH (multiple selections allowed; denominator = 262 respondents).

Assessment of hemorrhage burden and extension:

Routine hematoma volume quantification was inconsistent, with 41.5% of respondents reporting that they did not routinely quantify hematoma volume (Table 2). For quantification, the ABC/2 method (or reporting the three dimensions of the lesion) was the most common approach (40.7%), whereas automated software measurements were reported by 14 respondents (5.3%). Assessment of perihematomal edema was predominantly qualitative, with 60.1% providing a visual estimate and 27.0% commenting only when edema was radiologically significant. On the other hand, reporting of intraventricular hemorrhage and/or hydrocephalus was nearly universal, with 95.5% of respondents indicating they almost always included these findings in their radiology reports.

Reporting radiological predictors of hematoma expansion:

Most respondents reported almost always evaluating CTA or post-contrast CT examinations for the presence of a spot-sign (85.6%) and documenting a positive spot-sign when present (91.3%) (Table 3). In contrast, reporting the absence of a spot-sign was uncommon, with only 18.8% almost always documenting a negative finding and 64.7% rarely doing so. Assessment of non-contrast CT markers predictive of hematoma expansion (such as the swirl sign) was less common than CTA spot-sign (Table 3): 43.0% of respondents almost always evaluated and reported these markers when present, and 12.5% almost always documented both their presence and absence. when neuroradiologists identified imaging predictors of hematoma expansion, 48.1% deferred to the clinical team to determine the need for or timing of follow-up imaging (Figure 2); and less than half recommend follow-up scans within 6 hours (36.5%) or 6–12 hours (13.9%).

Table 3.

Reporting of spot sign on CTA and radiological predictor signs of hematoma expansion on non-contrast head CT.

	Response	Number (%)
Evaluating CTA or post-contrast CT for spot-sign (n=264)	Rarely (0–25%)	7 (2.7%)
	Occasionally (26–50%)	13 (4.9%)
	Frequently (51–75%)	18 (6.8%)
	Almost always (76–100%)	226 (85.6%)
Reporting a positive spot-sign when identified (n=264)	Rarely (0–25%)	5 (1.9%)
	Occasionally (26–50%)	5 (1.9%)
	Frequently (51–75%)	13 (4.9%)
	Almost always (76–100%)	241 (91.3%)
Reporting a negative (or absence of) spot-sign when not identified (n=266)	Rarely (0–25%)	172 (64.7%)
	Occasionally (26–50%)	22 (8.3%)
	Frequently (51–75%)	22 (8.3%)
	Almost always (76–100%)	50 (18.8%)
Evaluation and reporting of predictors signs of hematoma expansion on non-contrast head CT (n=263)	Not checked or reported	1 (0.4%)
	Rarely (0–25%) check and report only if positive	49 (18.5%)
	Occasionally (26–50%) check and report only if positive	32 (12.1%)
	Frequently (51–75%) check and report only if positive	34 (12.8%)
	Almost always (76–100%) check and report only if positive	114 (43.0%)
	Rarely (0–25%) check and report either the presence or absence	0
	Occasionally (26–50%) check and report either the presence or absence	0
	Frequently (51–75%) check and report either the presence or absence	2 (0.8%)
	Almost always (76–100%) check report either the presence or absence	33 12.5%)

Percentages are of valid responses for each item listed in the left column.

Figure 2.

Follow-up imaging recommendation when hematoma expansion is suspected (266 respondents).

Hemorrhage localization:

Confidence in assigning a primary ICH location when hematomas crossed anatomical boundaries was moderate overall, with 25.6% of respondents reporting near-certainty and 41.4% reporting being very confident (Table 4). The most common approach for determining the primary location for such ICH lesions, was the center of the hematoma (40.7%), followed by the presumed site of ICH origin (33.1%). Most respondents (78.9%) used a multi-regional description (such as “thalamocapsular ICH with lobar extension”) when reporting mixed-location hemorrhages (Table 4).

Table 4.

Localization of hematoma.

	Response	Number (%)
Confidence in assigning the location when ICH crosses anatomical borders (n=266)	Slightly confident (0–25%)	5 (1.9%)
	Moderately confident (26–50%)	83 (31.2%)
	Very confident (51–75%)	110 (41.4%)
	Near certain (76–100%)	68 (25.6%)
Determining the primary location of an ICH traversing across multiple regions (n=263)	Center of hematoma mass	107 (40.7%)
	Presumed origin of bleeding	87 (33.1%)
	Region with largest volume	38 (14.4%)
	Involvement of deep structures (thalamus, basal ganglia)	23 (8.7%)
	Combination of above (gestalt)	8 (3.0%)
Describing mixed-location hematomas (n=266)	Single dominant region (e.g., “deep thalamic ICH”)	15 (5.6%)
	Multi-regional description (e.g. “basal ganglia with lobar extension")	210 (78.9%)
	Qualitative narrative only (no specific label)	40 (15.0%)
	Combination of location and extension	1 (0.4%)

Percentages are of valid responses for each item listed in the left column.

Assessment of ICH etiology:

Confidence in distinguishing hypertensive, CAA, and anticoagulant-related ICH based on imaging alone was modest, with only 6.0% of respondents reporting near certainty, whereas 61.6% were only moderately or slightly confident (Table 5). On non-contrast CT, commenting on a potential CAA etiology based on hemorrhage location or the presence of subarachnoid hemorrhage (SAH) – analogous to the Edinburgh criteria – was variable, with only 20.0% reporting they almost always included such an assessment in their report. On MRI, the Boston Criteria v2.0 were the most commonly used framework for suggesting CAA (45.8%), followed by the presence of lobar microbleeds and/or cortical superficial siderosis (39.0%); however, 11.7% relied on general radiologic impression without specific criteria. When MRI findings were suggestive of CAA, 67.3% of respondents reported that they almost always included CAA as a potential etiology in their report.

Table 5.

Neuroimaging assessment of the ICH etiology.

	Response	Number (%)
Confidence in differentiating hypertensive, CAA, or anticoagulant-related ICH (n=266)	Slightly confident (0–25%)	40 (15.0%)
	Moderately confident (26–50%)	124 (46.6%)
	Very confident (51–75%)	86 (32.3%)
	Near certain (76–100%)	16 (6.0%)
Commenting on potential CAA-related ICH based on location or SAH presence on head CT (n=265)	Rarely (0–25%)	66 (24.9%)
	Occasionally (26–50%)	68 (25.7%)
	Frequently (51–75%)	78 (29.4%)
	Almost always (76–100%)	53 (20.0%)
Criteria for suggesting CAA etiology on brain MR (n=264)	Boston Criteria v2.0	121 (45.8%)
	Presence of lobar microbleeds and/or cortical superficial siderosis	103 (39.0%)
	Presence of lobar microbleeds only	9 (3.4%)
	No specific criteria (general radiologic impression/gestalt)	31 (11.7%)
Reporting CAA as a potential etiology when MRI criteria are present (n=266)	Rarely (0–25%)	10 (3.8%)
	Occasionally (26–50%)	19 (7.1%)
	Frequently (51–75%)	58 (21.8%)
	Almost always (76–100%)	179 (67.3%)

Percentages are of valid responses for each item listed in the left column.

Discussion

Our survey of 267 neuroradiologists, predominantly practicing in academic institutions and comprehensive stroke centers, identified several actionable gaps between current ICH imaging and reporting practices with emerging evidence, evolving best practices, and selected evidence-based recommendations from the AHA/ASA guidelines.^{3, 4, 6–10} The finding that more than 97% of respondents reported 24/7 emergency head CT interpretation coverage is reassuring and reflects widespread adoption of timely ICH detection and communication, the most fundamental component of acute ICH imaging and care. However, there remain opportunities to improve ICH imaging workflows and reporting practices, particularly with respect to hematoma volume quantification, CTA (or MRA) evaluation of ICH, incorporation of radiological predictors of hematoma expansion, follow-up imaging recommendations, and more consistent localization and etiologic classification.

The finding that 42% of respondents do not routinely quantify hematoma volume represents a clinically important practice gap with direct implications for guideline adherence and quality metrics. The Joint Commission requires documentation of the baseline “ICH score” within 6 hours of hospital arrival for patients with spontaneous ICH,¹² and two of its components – hematoma volume (>30 mL) and infratentorial location – are determined by neuroimaging.¹³ Documentation of baseline ICH severity is also a 2024 AHA/ASA Class 1 quality performance measure.¹⁰ Beyond incorporation in the “ICH score”, hematoma volume directly informs treatment decisions, including eligibility for minimally invasive hematoma evacuation of (30–80 mL) lobar,⁴ decompressive craniectomy of 30–100mL deep,⁵ and urgent surgical evacuation of (≥15 mL) cerebellar hemorrhage.³ Among respondents who quantified hematoma volume, the ABC/2 method was the most commonly used approach, consistent with the technique used in most ICH clinical trials.^{4, 8, 14, 15} Moreover, ABC/2 volume can be readily estimated from three orthogonal lesion dimensions reported on CT.¹⁵ As automated volumetric quantification software becomes increasingly available, incorporation of precise ICH volume measurements into routine clinical reports should become even more feasible. Overall, our findings identify a readily addressable gap in neuroradiology reporting that could be improved through routine use of the ABC/2 method, three-dimensional measurements, or automated volumetric tools.

In light of emerging treatment paradigms, incorporation of CTA/MRA into the standard ICH imaging workflow remains an important opportunity for improvement. Only 45% of respondents reported that CTA is routinely integrated into their institutional ICH imaging protocol, and nearly one-third reported that they only occasionally or rarely recommend CTA following non-contrast CT. This is notable because the 2022 AHA/ASA guidelines recommend vascular imaging to evaluate for macrovascular causes of hemorrhage in select patients with lobar ICH and in younger patients or those without hypertension.³ However, diagnostic vascular imaging for both lobar and non-lobar ICH is among 2024 AHA/ASA Performance Measures.¹⁰ Beyond identifying an underlying vascular etiology, CTA can detect active contrast extravasation (the spot-sign), a well-established predictor of hematoma expansion that may influence clinical management by prompting earlier follow-up CT (typically within 3 hours) and consideration of more aggressive hemostatic therapy in patients receiving anticoagulant or antiplatelet medications. More recently, CTA has gained additional clinical relevance following the FASTEST randomized trial, which suggested that patients with spot-sign–positive ICH may benefit from ultra-early hemostatic therapy.⁸ Although broader integration of CTA into routine ICH evaluation extends beyond 2022 guideline recommendations for all patients, it represents an important opportunity to adopt emerging treatment strategies and align with the 2024 ICH Performance and Quality Measures.¹⁰

Hematoma expansion is a major determinant of neurological deterioration, mortality, and functional outcome after ICH.³ Accordingly, the 2022 AHA/ASA guidelines include a Class IIb recommendation to assess the CTA spot-sign, the most extensively validated imaging predictor of hematoma expansion,¹⁶ as well as non-contrast CT markers – such as the swirl, black hole, blend, island, and satellite signs – which may serve as alternatives or adjuncts when CTA is unavailable. Our survey demonstrated excellent adoption of CTA spot-sign evaluation, with 86% of respondents almost always assessing for a spot-sign and 91% reporting positive findings. However, only 19% almost always documented the absence of a spot-sign, making it difficult for clinicians to distinguish between a true negative examination and a finding that was simply not assessed. In contrast, adoption of non-contrast CT markers of hematoma expansion was substantially lower, with only 43% of respondents almost always evaluating and reporting these findings despite consensus recommendations from the International NCCT ICH Study Group.¹⁷ Our findings highlight an opportunity to improve neuroradiology reporting by more consistently assessing and documenting established non-contrast CT markers of hematoma expansion. Although these markers exhibit greater interobserver variability than the CTA spot-sign, their recognition is increasingly important given their role in risk stratification and their potential to identify patients most likely to benefit from emerging hemostatic therapies.

Regarding ICH location, current management guidelines and severity metrics (i.e. the ICH score), distinguish only infratentorial from supratentorial hemorrhage.^{3, 13} However, since minimally invasive evacuation appears to be selectively effective in lobar ICH,⁴ more precise lobar versus deep localization is becoming increasingly important for risk stratification and treatment selection. Our survey highlights the challenges of assigning a dominant location when hemorrhages cross anatomical boundaries, with considerable variability in both confidence and localization strategies among neuroradiologists. Although automated atlas-based methods may ultimately standardize this process,¹⁸ our findings underscore the need for a consensus standardization of ICH localization in neuroradiology community, particularly for distinguishing lobar from non-lobar hemorrhages.

Our findings also identified opportunities to improve neuroradiologists’ evaluation and reporting of ICH etiology, particularly CAA-related hemorrhage. Only 38% of respondents reported being very confident or nearly certain in distinguishing hypertensive from CAA-related ICH based on imaging alone. Although the Boston Criteria version 2.0 are the current standard for MRI-based diagnosis of possible or probable CAA,¹⁹ only 46% of respondents reported routinely using these criteria, and approximately one-half rarely or only occasionally applied CT-based (Edinburgh)²⁰ criteria to suggest possible CAA. Notably, once imaging findings were considered suggestive of CAA, 67% of respondents almost always included CAA as a potential etiology in their report. These findings suggest that while neuroradiologists recognize the importance of reporting CAA when suspected, there is substantial variability in the systematic evaluation of imaging findings that support its diagnosis. As CAA-related ICH are associated with higher risk of recurrence,¹¹ and may require different anticoagulation strategies when needed,²¹ greater familiarity with established imaging criteria may improve diagnostic consistency and reporting.

Our survey also underscores the potential for neuroradiologists to play a more proactive role in guiding imaging follow-up for patients with acute ICH. Notably, 41% of respondents reported that they do not routinely recommend follow-up imaging unless requested by the treating team, and 48% indicated that they defer recommendations for repeat imaging even when imaging features suggest an increased risk of hematoma expansion. Although decisions regarding repeat imaging are ultimately multidisciplinary and depend on the patient’s clinical course, neuroradiologists are uniquely positioned to recommend evidence-based imaging follow-up, particularly during the initial evaluation when ICH care often begins at community hospitals.

This study has several limitations. Our investigator-developed questionnaire was not previously validated and did not undergo a formal Delphi consensus process, which may affect the reliability and reproducibility. As with all survey-based studies, the results reflect self-reported practices and are therefore subject to response and recall bias. Moreover, the 267 neuroradiologists who participated in this survey represents fewer than 8% of practicing neuroradiologists in the United States.²² Furthermore, respondents were predominantly from academic institutions and comprehensive stroke centers, which may limit the generalizability of our findings to community practice settings. The questionnaire also did not distinguish head-only from combined head-and-neck CTA; therefore, the reported CTA practices should be interpreted as general practice patterns rather than estimates of specific CTA acquisition protocols. The questionnaire did not stratify follow-up imaging recommendations by hemorrhage pattern or suspected etiology; therefore, these findings should not be interpreted as scenario-specific imaging recommendations. The questionnaire did not assess respondents’ familiarity with non-contrast CT hematoma-expansion signs, which may have influenced interpretation of questions addressing these imaging markers. Because the survey was anonymous, multiple respondents may have originated from the same institution, precluding adjustment for institution-level clustering and potentially overrepresenting local practice patterns (e.g., routine CTA protocols). Consequently, we intentionally limited our analysis to descriptive statistics and did not examine associations between practice setting and reporting behavior. Thus, our findings should be interpreted as an overview of contemporary neuroradiology practice patterns rather than a precise estimate of national practice.

Conclusions

This national survey identified several potentially actionable gaps in the neuroimaging evaluation and reporting of spontaneous ICH. Readily addressable opportunities include routine hematoma volume quantification (or reporting of three orthogonal hematoma dimensions), routine incorporation of CTA to evaluate for macrovascular etiologies and spot-sign, more consistent assessment and reporting of non-contrast CT predictors of hematoma expansion, evidence-based recommendations for follow-up imaging, and standardized evaluation and reporting of CAA on MRI using the Boston Criteria version 2.0. Our findings also highlight the need for evidence-based consensus approach to localizing ICH that traverse anatomical boundaries, particularly to improve distinction between lobar and non-lobar ICH, an issue that is likely to become increasingly important for surgical patient selection. Finally, broader dissemination of ICH imaging and reporting best practices to community hospitals and private practice settings, which were underrepresented in this survey, could extend these benefits to a larger population of patients with ICH.

Summary:

Previous Literature:

Contemporary ICH management increasingly relies on neuroimaging to guide time-sensitive treatment. AHA/ASA guidelines recommend rapid CT or MRI, vascular imaging in selected patients, hematoma volume assessment, evaluation of the CTA spot-sign and non-contrast CT markers of hematoma expansion, and MRI-based assessment of underlying etiologies such as cerebral amyloid angiopathy. However, how consistently neuroradiologists incorporate these recommendations into routine imaging and reporting practice has not been systematically characterized.

Key Findings:

Among 267 respondents, important practice gaps included inconsistent hematoma volume quantification, limited routine CTA use, variable assessment of non-contrast CT markers of hematoma expansion, and modest confidence in hemorrhage localization and etiologic classification.

Knowledge Advancement:

This survey identifies actionable opportunities to standardize ICH imaging and reporting, including routine hematoma volume measurement, vascular imaging, documentation of hematoma expansion markers, evidence-based follow-up recommendations, and more consistent etiologic assessment.

Abbreviations:

AHA

American Heart Association

ASA

American Stroke Association

CAA

Cerebral Amyloid Angiopathy

ICH

Intracerebral Hemorrhage