Search Results for: L'imaging nella urolitiasi

Stroke, Volume 56, Issue Suppl_1, Page AWP6-AWP6, February 1, 2025. Background and Purpose:Whether imaging markers of cerebral small vessel disease on computed tomography (CT-CSVD) relates to early clinical outcomes after intravenous thrombolysis for acute ischemic stroke remains not well understood. This multicenter retrospective study aims to investigate the association between baseline CT-CSVD score and early clinical outcomes in acute ischemic stroke patients who received intravenous thrombolysis.Methods:Individual CT-CSVD imaging markers and total score were assessed based on the following conditions: severe white matter lucencies (grade 2), two or more lacunes, and severe central or cortical atrophy (grade 2), with scores ranging from 0 to 3. The primary outcome was functional independence, defined as a modified Rankin Scale score of 0-2 at discharge. The secondary outcome was early neurological deterioration, defined as a 2-point or greater decline on the NIHSS within the first 24h of admission. Multivariate logistic regression analyses were used to examine the associations between CT-CSVD scores and outcomes.Results:Of 920 eligible patients (median age 68, male 65.8%), 9.9% (91/920) experienced early neurological deterioration, and 62.1% (545/877) achieved functional independence at discharge. Patients with increasing CT-CSVD scores were older (p

Stroke, Volume 56, Issue Suppl_1, Page AWP198-AWP198, February 1, 2025. Introduction:Vertebrobasilar artery dissection (VBD) results from blood entering a tear in the intima of the vertebral artery (VA), basilar artery, and their branches. T1-Sampling Perfection with Application optimized Contrast using different flip angle Evolution (T1-SPACE) is valuable for detecting intramural hematoma (IMH). Despite the potential of susceptibility-weighted imaging (SWI) for IMH detection, its clinical adoption remains limited. This study aimed to clarify the usefulness of SWI and T1-SPACE in detecting IMH in VBD.Methods:Consecutive patients admitted with symptomatic VBD within 28 days of initial symptom onset between November 2014 and September 2023 were retrospectively screened from the Jikei University School of Medicine Stroke Registry. The inclusion criteria were (1) definitive VBD diagnosis based on the Spontaneous Cervicocephalic Arterial Dissections Study criteria, and (2) performance of both SWI and T1-SPACE simultaneously. IMH on SWI was identified by (1) eccentric or concentric areas of hypointensity adjacent to the vessel wall, (2) hyperintensity on phase maps corresponding to hypointensity on SWI, and (3) absence of calcification on CT. We compared the IMH detectability between SWI and T1-SPACE.Results:The study included 74 patients (52 males, median age 45). The 65 patients detected IMH on SWI, the 66 patients detected on T1-SPACE; IMH detection rates were equivalent (88% vs. 89%; p=1.00). T1-SPACE had a higher detection rate in VA dissection compared to PICA dissection (97 vs 14%, p

Stroke, Volume 56, Issue Suppl_1, Page AWP190-AWP190, February 1, 2025. Background:Predicting recanalization in acute ischemic stroke (AIS) patients undergoing mechanical thrombectomy (MT) may help in optimizing treatment strategies but remains a critical challenge. Patients presenting intracranial atherosclerosis-related large vessel occlusion (ICAS-LVO) may benefit from performing an early bailout strategy after failed MT with conventional devices. Current predictive models often fail to effectively integrate complex interactions between clinical data and neuroimaging features. We aimed to explore predictive models for ICAS-LVO after an unsuccessful first pass, combining neuroimaging and clinical data.Methods:This retrospective, single-center study of patients presenting anterior LVO stroke from 2017 to 2023. ICAS-LVO was defined by fial angiographic stenosis >50% or failed MT (eTICI0-2a) with no evidence of other underlying etiology (i.e cardioembolic). Neuroimaging data included raw non-contrast CT and CTA volumes, automatic vascular segmentation, and CTP output values. Clinical data, such as patient demographics, stroke severity, and comorbidities, were incorporated into the model. A Dynamic Affine Feature Map Transform (DAFT) mechanism was used to integrate imaging data, encoded with a convolutional neural network, with clinical variables. A comparative analysis including several image-based deep learning models and known machine learning models trained on clinical data was performed. Data were split into 80% for training with four-fold cross-validation and 20% for testing.Results:A total of 262 patients (median age 81 IQR 69-88, 61% women; 11.83% with ICAS-LVO) with anterior LVO and failed MT first attempt (eTICI 0-2a) were included. Training sample was of 212 patients (12.73% with ICAS-LVO). A consecutive sample of 50 patients (8% with ICAS-LVO) was reserved for testing. The best-performing model achieved an AUC of 0.73 ± 0.10 (mean ± std), and a F1 of 0.44 ± 0.15 in validation and an AUC of 0.64 ± 0.12, and a F1 of 0.21 ± 0.09 on internal testing.Conclusions:Although effective prediction of ICAS-LVO remains challenging, models using multi-modal data were superior to those using imaging or clinical data alone emphasizing the importance of combining multiple sources of data to enhance predictive capabilities. This model could be useful to rapidly identify patients with suspected ICAS-LVO prior to MT and optimize endovascular treatment strategy after the initial attempt.

Stroke, Volume 56, Issue Suppl_1, Page AWMP101-AWMP101, February 1, 2025. Introduction:The ARCADIA trial (Apixaban to Prevent Recurrence After Cryptogenic Stroke in Patients With Atrial Cardiopathy) found no benefit of anticoagulation for secondary stroke prevention in patients with cryptogenic stroke and evidence of atrial cardiopathy. It remains unclear if the biomarkers used in the trial reliably identified atrial cardiopathy. We examined the association between biomarkers of atrial cardiopathy and acute brain infarction in multiple arterial territories, an imaging signature of cardioembolic stroke.Hypothesis:Biomarkers of atrial cardiomyopathy are associated with acute brain infarction in multiple arterial territories.Methods:The ARCADIA trial screened patients with cryptogenic stroke for atrial cardiopathy at 185 centers in the U.S. and Canada. Investigators were asked to record the presence and topography of acute brain infarction on baseline imaging. Multi-territorial infarction was defined as acute infarction in at least two of the left middle cerebral artery (MCA), right MCA, and posterior circulation. P-wave terminal force in ECG lead V1 (PTFV1) and left atrial dimension index (LADI) were modeled as continuous variables, whereas N-terminal pro-B-type natriuretic peptide (NT-proBNP) was log-transformed. Relative risk regression was used to examine the association between atrial cardiopathy biomarkers and multi-territorial infarction.Results:Of 3,745 patients enrolled in ARCADIA, 3,301 had available data on atrial cardiomyopathy biomarkers and the topography of acute infarctions seen on baseline imaging. Of these 3,301 patients, 452 (13.7%) had multi-territory brain infarction. We found no association with multi-territory brain infarction for ln(NT-proBNP) (OR per SD, 1.06; 95% CI, 0.97-1.16), PTFV1 (OR per SD, 1.06; 95% CI, 0.98-1.15), or LADI (OR per SD, 0.95; 95% CI, 0.86-1.05). Patients who met criteria for atrial cardiopathy and were randomized into the treatment phase of the study had a similar likelihood of multi-territorial infarction (13.4%) compared with those who were not eligible for randomization (13.8%) (P = 0.77).Conclusions:The atrial cardiopathy biomarkers used in the ARCADIA trial were not associated with patterns of brain infarction suggestive of a cardioembolic source.

Stroke, Volume 56, Issue Suppl_1, Page ATMP49-ATMP49, February 1, 2025. Background and Objective:A stable perfusion pattern may guarantee a good clinical outcome in acute stroke patients. This study aimed to explore the differences of perfusion patterns over the acute phase in acute stroke patients with different etiologies.Methods:This study included acute ischemic stroke patients who arrived at Huashan Hospital, Fudan University within 24 hours of last known normal. All patients underwent baseline multimodal CT scan (including CT angiography [CTA] and perfusion imaging), and had confirmed large vessel occlusion or severe stenosis of anterior circulation. Recanalization status was evaluated through follow-up perfusion or angiography imaging. Final infarct volume (FIV) was assessed by follow-up CT or magnetic resonance imaging within 7 days after arrival. Penumbral stability was assessed using the Perfusion-Infarct Index (PFI)，which was defined as 1 – (FIV – baseline infarct core) / baseline penumbra. For patients without recanalization, a stable penumbral pattern was defined PFI > 0.9. Kaplan-Meier survival curve analysis was used to explore the difference of PFI between patients with cardioembolic (CE) and large artery atherosclerotic (LAA) etiology during the acute phase. For patients without recanalization, stepwise logistic regression analysis was conducted to identify independent predictors of a stable penumbral pattern.Results:A total of 250 acute ischemic stroke patients were included, with 160 classified as LAA-related stroke and 90 as CE-related stroke. Kaplan-Meier survival curve analysis revealed that LAA-related stroke patients had a higher PFI value compared to CE-related stroke patients throughout the acute phase (P=0.02). Of the 88 patients who did not achieve recanalization, patients with LAA-related stroke were more likely to have a stable penumbral pattern compared with patients with CE-related stroke (39% vs. 10.3%, P=0.01). Stepwise logistc regression analysis demonstrated that baseline NIHSS ≤8 (OR: 8.8, 95% CI: 2.9 – 26.4, P

Stroke, Volume 56, Issue Suppl_1, Page AWP193-AWP193, February 1, 2025. Introduction:Intracranial atherosclerotic disease (ICAD) is a common cause of primary and secondary stroke. Understanding plaque evolution can help in risk stratification and in selecting proper therapies.Hypothesis:In this study, we hypothesized that vessel wall radiomics and hemodynamic features correlate with longitudinal change in plaque volume.Methods:Patients with ICAD and history of cerebral ischemia were included. Time-of-flight MRA, high-resolution vessel wall imaging (proton density-weighted, PD, and contrast-enhanced T1-weighted, T1CE) and phase-contrast MR-based volumetric flow (NOVA) were collected at multiple time points. MRA was used to segment lumen and perform computational fluid dynamics (CFD). NOVA flow rate used as boundary conditions. Plaques were segmented on PD images from each time point. PD and T1CE scans were co-registered and radiomics features (RFs) were extracted from the first time point from both image types. Plaques were compared between consecutive time points and classified as progressing, stable, or regressing, based on volume change (those

Stroke, Volume 56, Issue Suppl_1, Page ATP303-ATP303, February 1, 2025. Background:Prior studies indicate a relationship between migraine and MACE. Here we assess whether brain white matter hyperintensities (WMH), which are a common imaging finding in patient with migraine, contribute to the risk of MACE above demographics and common risk factors for MACE.Methods:60,454 patients, ages 18-89, with a ICD-9 or ICD-10 migraine diagnosis code in 2010 or later were identified from the Mayo Clinic electronic health record. Only patients who were seen for migraine in the Neurology Department were included. Patients with a migraine diagnosis who did not have MACE were included only if they had at least two visits at Mayo Clinic during five years. Only patients with sex and race information were included. The final cohort included 577 migraine patients with and 598 migraine patients without MACE. Presence of WMH was determined from radiology notes. Individuals without a brain MRI were assumed not to have WMH. A logistic regression model that included sex, race, known lifetime MACE risk factors (atrial fibrillation, diabetes, hypertension, hyperlipidemia, tobacco use) and WMH as independent variables to predict MACE outcome was fit.Results:Significant factors that increased the risk of MACE in individuals with migraine included being Black or African American (adjusted OR: 2.9, 95% CI: 1.24-6.82, p = 0.014), presence of atrial fibrillation (adj. OR: 1.63, 95% CI: 1.23-2.17, p < 0.001), diabetes (adj. OR: 1.34, 95% CI: 1.02-1.75, p = 0.036), hypertension (adj. OR: 1.9, 95% CI: 1.39-2.6, p < 0.001), tobacco use (adj. OR: 1.66, 95% CI: 1.29-2.14, p < 0.001), and the presence of WMH (adj. OR: 1.43, 95% CI: 1.1-1.87, p = 0.008). Hyperlipidemia showed a marginal association (adj. OR: 1.34, 95% CI: 0.99-1.81, p = 0.061), while other variables such as sex and other racial/ethnic groups did not significantly alter the risk of MACE outcome.Discussion:Results indicate that African American race and presence of WMH in addition to common comorbidities independently increase the risk of MACE outcome.

Stroke, Volume 56, Issue Suppl_1, Page ATMP18-ATMP18, February 1, 2025. Background and Issues:Timely identification and intervention are critical for positive outcomes in acute intracerebral hemorrhage (ICH) patients. The 2022 Guideline for the Management of Patients With Spontaneous Intracerebral Hemorrhage: A Guideline From the American Heart Association/American Stroke Association states that we should control blood pressure, reverse anticoagulation, and get the patient to a specialized inpatient stroke unit. Unlike multiple clearly defined target goals for ischemic stroke, there are no established measurable goals for ICH. The absence of specific target times for ICH leads to delays in identification and treatment, resulting in poor outcomes.Purpose:This project aimed to rapidly identify and aggressively manage ICH patients with clearly defined treatment times to reduce mortality and 30 day readmissions.Methods:The implementation of measurable goals for ICH in the Emergency Department (ED) was an approach to improve ICH patient outcomes by facilitating faster identification and treatment. The stroke team reviewed existing performance data and outcomes to determine reasonable times for completing the following interventions: Cat Scan imaging within 30 minutes of being ordered, blood pressure medications given within 30 minutes of diagnosis, reversing anticoagulation within 45 minutes of diagnosis, and moving patients to the Neuro ICU within 90 minutes of diagnosis. The ED team was educated on the new performance goals. The stroke team collected, analyzed, and reported the performance metrics of the ED team throughout implementation. Outcomes were measured using data for when the goals were being met versus not being met.Results:Outcomes were measured from 111 ICH patients from 2022-2023. When treatment time goals were not met, the mortality rate for ICH patients was 18.42% with an observed/expected (O/E) ratio of 1.25. The 30 day readmission rate was 12.9%. When treatment goals were met, the mortality rate decreased to 12.68% with an O/E ratio of 0.86. 30 day readmission rate dropped to 3.23%Conclusions:Having measurable goals for ICH has demonstrated a significant improvement in outcomes. Establishing clear time targets for getting imaging, initiating blood pressure control, reversing anticoagulation, and admission to Neuro ICU led to a reduction in mortality and 30 day readmissions. It is recommended to use clearly defined time goals to ensure prompt intervention and improved prognosis for patients with ICH.

Stroke, Volume 56, Issue Suppl_1, Page AWMP105-AWMP105, February 1, 2025. Introduction:Accurate prediction of the risk of ischemic stroke (IS) is vital for prevention and would be aided by multimodal biomarkers integrating genetic, clinical, and functional data. The role of imaging and EKG based atrial measurements, other than atrial fibrillation (AF), in IS prediction is uncertain and many strokes remain cryptogenic despite extensive work-up. As an exploratory step to improve stroke evaluation by including atrial traits, we developed a novel multimodal deep learning model integrating demographic and clinical variables with atrial phenotypic and genotypic data.Methods:We collected individuals from UK Biobank (UKBB) and defined ischemic stroke (IS) by the UKBB Algorithmically Defined Outcome (ADO). We developed a multimodal multi-layer perceptron with late fusion (MMLP-LF) model to predict whether a subject has IS by integrating five data modalities from UKBB: 1) MRI and EKG derived atrial traits, 2) lead genetic variants (P

Stroke, Volume 56, Issue Suppl_1, Page ATMP24-ATMP24, February 1, 2025. Background and Purpose:The Stroke Response Team (SRT) is comprised of a critical care RN with neuro experience and a Neuro ICU APP. The SRT RN acts as a critical link in the care continuum for stroke patients by providing consistent ICU-level care from initial entry into the ED through transitions to procedural areas and the ICU. The SRT resulted from the need to increase patient safety, enhance continuity of care, and improve clinical outcomes. Additional aims are to improve compliance with core measures and standardization of patient care. Data collection is ongoing. Before implementation, charting metrics were at

Stroke, Volume 56, Issue Suppl_1, Page AWP194-AWP194, February 1, 2025. Introduction:A semi-automated method that compares voxel density with the contra-lateral hemisphere to generate ratio, or relative Non-Contrast CT (rNCCT) maps for identifying hypodensity changes was developed. In addition to being sensitive to stroke related hypodensities, these maps are also sensitive to motion artefacts and naturally occurring asymmetry in densities across hemispheres. We assessed the value of a deep-learning based model to segment and remove these artefacts and for identifying ischemic core of baseline NCCT.Methods:We included data from 268 acute ischemic stroke patients with a large vessel occlusion from the ongoing CT perfusion to Predict Response to Recanalization in Ischemic Stroke Project 2 study. NCCT scans acquired at the primary stroke center were used to create rNCCT maps. These maps detect regions with at least 1% relative hypodensity difference compared to the contralateral region. A trained observer who had insight of arterial occlusion location manually annotated artefacts. We trained ano new UNetusing the NCCT, rNCCT, and flipped NCCT images to detect artefacts from the rNCCT maps. To assess the extent to which our model falsely identified ischemic regions as artefact, we determined the overlap between the automatically segmented artefact on the rNCCT map and the manually segmented ischemic core on diffusion-weighted imaging (DWI) acquired at the comprehensive stroke center before treatment.Results:The best performing model was the ensemble of the five cross-validation folds of 3d low- and high-resolution models based on dice similarity coefficient. Figure 1 provides an example of our model’s artefact segmentation and the processed rNCCT map after artefact removal. For the 54 patients (20% of study population) in our test set, our model achieved a median Dice similarity coefficient of 0.95 (IQR: 0.91-0.97) and a median false positive volume of 6.1 (3.2-11) ml. In the 30 patients with available DWI scans, 30% of patients had any overlap ( >=1 voxel) between the segmented artefact and DWI ischemic core with a median overlap volume of 0.69 (IQR: 0.32-2.3) ml.Conclusion:We demonstrate the use of a deep-learning based model to automatically segment artefacts from rNCCT maps. Our model circumvents time-invasive manual removal of artefacts from the rNCCT map and thereby simplifies segmentation of the ischemic core on baseline NCCT. Validation with external datasets is necessary before use in routine stroke evaluation.

Stroke, Volume 56, Issue Suppl_1, Page AWMP109-AWMP109, February 1, 2025. Introduction:During acute ischemic stroke, energy depletion leads to a rise in creatine (Cr) as a buffer, increased extracellular glutamate from impaired neurotransmitter transport, and a pH drop due to lactic acid buildup. Understanding these changes is crucial for timely intervention, yet no current method captures all these alterations simultaneously. CEST, a novel MRI technique, non-invasively maps metabolites with high sensitivity, providing information on both concentration and pH. Our group recently demonstrated guanidino CEST (GuanCEST) at 3T, reflecting Cr levels, while amine CEST (amineCEST) at 9.4T likely indicates glutamate, and amide CEST (amideCEST) correlates with pH. We aim to use these techniques to investigate metabolic, neurotransmitter, and pH changes in two mouse models of middle cerebral artery occlusion (MCAO).Methods:Ten male C57BL/6 mice (aged 3–6 months) were used for two stroke models: permanent MCAO (pMCAO, n=4) and transient MCAO (tMCAO, n=4). MRI scans were conducted at 9.4T and 3T. Diffusion-weighted imaging identified stroke lesions, followed by T1and T2mapping on the selected slice. CEST scans were performed with a 2s saturation time across B1values ranging from 0.4 to 3.0 μT. We utilized Polynomial and Lorentzian Line-shape Fitting (PLOF) to simultaneously extract GuanCEST, amineCEST, and amideCEST from the CEST spectrum at each pixel, generating corresponding CEST maps. The average CEST values in the lesion and contralateral hemisphere were analyzed.Results:At a B1of 0.4 μT, GuanCEST (Fig. 1, red line) increased by 1.01±0.19% in pMCAO compared to the contralateral hemisphere but decreased by 0.32±0.27% in tMCAO, indicating a greater Cr rise in pMCAO. At higher B1, Cr effects diminished while pH effects increased. When B1exceeded 0.8 μT, amineCEST increased by 3.86±0.42% in tMCAO, nearly four times the rise in pMCAO (1.09±0.26%), possibly reflecting neurotransmitter changes due to cell membrane polarization and energy depletion. At 1.6 μT, amideCEST decreased by 0.53±0.05% in pMCAO but remained stable in tMCAO (0.49±0.48%), suggesting greater tissue acidification in pMCAO. Similar trends were observed at 3T, except amineCEST was undetectable.Conclusion:CEST MRI is a non-invasive technique capable of mapping metabolite, neurotransmitter, and pH changes in the stroke-affected brain, with strong potential for clinical translation.

Stroke, Volume 56, Issue Suppl_1, Page AWMP107-AWMP107, February 1, 2025. Background:Early detection of carotid artery (CA) conditions such as atherosclerosis and dissection are important as they represent major preventable causes of stroke. While medical imaging equipment can track calcification and dissection, they are bulky, time-consuming and unsuitable for remote or continuous monitoring. In this study, we sought to analyze if a wearable, microchip seismometer patch can capture significant differences in the systolic and diastolic blood acoustic vibrational signature from the carotid arteries across various conditions.Methods:The multi-head StethX seismometer patches were placed on 63 patients seen at the Emory Stroke Clinic via medical tape near the left and right CA and the pulmonic region of the heart to capture the CA blood acoustic and heart sound vibrations. These patients had varying CA conditions (atherosclerosis, dissection, normal, etc.) confirmed by magnetic resonance angiography (MRA), computer tomography angiography (CTA) or carotid ultrasound within the past year and had their blood acoustic vibration signatures recorded for one minute. The signals were then converted into time-frequency representations (TFR) with color scale used as magnitude of signal using continuous wavelet transform to distinguish between healthy and unhealthy carotid arteries.Results:The TFR of the blood acoustic vibrations captured by the seismometer showed significant differences of as a function of varying CA conditions. For CA with less than 50% stenosis, two distinct peaks representing systolic and diastolic are present, with systolic being lower frequency than diastolic, as shown in the Figure. For CA stenosis of greater than 50%, specifically 70-99%, the diastolic peaks are in a lower frequency range compared to less than 50% stenosis. For CA dissection, the systolic and diastolic peaks both show a higher frequency range compared to patients with less than 50% stenosis.Conclusions:StethX’s wearable microchip seismometer patch captured distinct acoustic vibration patterns as shown via TFR in patients with varying CA conditions, including moderate to severe atherosclerotic stenosis and dissection, when compared to less than 50% CA stenosis. This enables the application of machine learning models that use the CA blood acoustics along with possibly pulse transit time using a third patch located on the heart pulmonic region as inputs to classify severity of carotid artery stenosis as a digital health screening solution.

Stroke, Volume 56, Issue Suppl_1, Page A80-A80, February 1, 2025. Introduction:Although flow diverters (FD) are a long-established treatment-option for brain aneurysms, the evidence of data associated with therapy success is still based on relatively small heterogeneous studies. Recent data from a large prospective registry with highly standardized, independent adjudication might provide more insights.Methods:The INSPIRE-A registry is a prospective, single arm study of brain aneurysms treated with commercially available Medtronic devices, including latest FD technology with surface modifications. The registry collects data from 70 centers in 21 countries. Results are adjudicated by an independent Imaging Core Laboratory and a Clinical Events Committee. The association of patient characteristics, aneurysm details, and procedural variables with aneurysm occlusion (Raymond–Roy occlusion classifications [RROC I]), and in-stent stenosis at 1 year follow-up (FU) were tested using multivariable logistic regression analysis.Results:786 patients were analyzed. The median age was 55 years (IQR, 46-64) and 79.6% (626) were female. Most frequently, aneurysms were in the anterior circulation (89.8 %, 706) with highest rates of ICA-Aneurysms (75.3%, 592) in the C6 (52.7%, 312/592). Median diameter and neck width of aneurysms were 7.2 mm (IQR, 5-11) and 6.5 mm (IQR, 4-10), respectively. 91.7% (719) of aneurysms were saccular. The Pipeline Flex with Shield technology (Medtronic, Irvine CA) device was deployed in 58% (456), the Pipeline Vantage with Shield technology (Medtronic, Irvine CA) device in 42% (330). Adjunctive coiling was used in 19.8% (156). At 1 year FU, RROC I was observed in 74.3% (584) of cases. In multivariable analysis, increasing age (OR, 0.97, 95% CI, 0.96-0.99; p

Stroke, Volume 56, Issue Suppl_1, Page AWP197-AWP197, February 1, 2025. Background:The hyperdense artery sign (HAS) in patients with large vessel occlusion (LVO) is associated with outcomes after ischemic stroke. Considering the labor-intensive nature of manual segmentation of HAS, we developed and validated an automated HAS segmentation algorithm on non-contrast brain CT (NCCT) images using a multicenter dataset with independent annotations by two experts.Methods:For the training dataset, we included patients with ischemic stroke undergoing concurrent NCCT and CT angiography between May 2011 and December 2022 from six stroke centers. The model was externally validated using a dataset from one stroke center. For the clinical validation dataset, a consecutive series of patients admitted within 24 hours of symptom onset were included between December 2020 and April 2023 from six stroke centers. The model was trained using a 2D U-Net algorithm with manual segmentation by two experts. We constructed models trained on datasets annotated individually by each expert, and an ensemble model using shuffled annotations from both experts. The performance of the models was compared using area under the receiver operating characteristics curve (AUROC), sensitivity, and specificity.Results:A total of 673, 365, and 774 patients were included in the training, external validation, and clinical validation datasets, respectively, with mean (SD) ages of 68.8 (13.2), 67.6 (13.4), and 68.8 (13.6) years and male frequencies of 55.0%, 59.5%, and 57.6%. The ensemble model achieved higher AUROC and sensitivity compared to the models trained on annotations from a single expert in the external validation dataset. In the clinical validation dataset, the ensemble model exhibited an AUROC of 0.846 (95% CI, 0.819–0.871), sensitivity of 76.8% (65.1–86.1%), and specificity of 88.5% (85.9–90.8%). The predicted volume of the clot was significantly correlated with infarct volume on follow-up diffusion-weighted imaging (r=0.42; p

Stroke, Volume 56, Issue Suppl_1, Page A97-A97, February 1, 2025. Objective:To evaluate a potential relationship between post-stroke cognitive impairment (PSCI) and a radiographic measure of glymphatic function after intracerebral hemorrhage (ICH).Introduction:PSCI has been less studied in patients with ICH. Here we aim to evaluate the role of glymphatic function after ICH and its association with PSCI. Glymphatic cleanup occurs in the perivascular space (PVS) formed by astroglial end-feet loosely surrounding small arteries and veins. Here we applied non-invasive diffusion tensor imaging (DTI) to measure changes in brain diffusion due to dynamics changes of interstitial fluid (ISF) and CSF along the PVS. The application of DTI-along the perivascular space index (DTI-ALPSI) has been validated to evaluate glymphatic function. It computes the diffusivity ratio between projection and association fibers oriented orthogonally with the medullary veins at the level of the lateral ventricle body (Fig-1A).Methods:We serially imaged 18 patients with deep ICH at 9.3±9.1 (V1) and 109±28 (V2) days of onset on a 3T MRI system. Cognitive assessment was obtained via MoCA scores. Fractional anisotropy (FA) and mean diffusivity (MD) maps were registered to the T1W and SWI images. Three regions of interest (ROI), remote from a lesion, in the association and projection fibers, orthogonal to the medullary veins at PVS were used to compute diffusivity (Fig-1C). Using the equation shown, the DTI-ALPSI was calculated. DTI-ALPSI = Mean (Dx proj, Dx assoc) / Mean (Dy proj, Dz assoc)Using 3D-Flair images, hematoma (HV) and edema (EV) volumes were segmented. The ipsilesional DTI-ALPSI was correlated with MoCA, HV, and NIHSS. Contralesional DTI-ALPSI was used as a control. A non-linear regression model was used for statistical analysis.Results:We enrolled 12M/6F with an average age of 49.3±13.3y. Compared to the control, the ipsilesional DTI-ALPSI was significantly decreased (p=0.036) at V2. Temporally the MoCA scores were significantly increased (19±8.5 to 23±6.0, p