Evaluation of Ischemic Penumbra in Stroke Patients Based on Deep Learning and Multimodal CT

In order to investigate the value of multimodal CT for quantitative assessment of collateral circulation, ischemic semidark zone, core infarct volume in patients with acute ischemic stroke (AIS), and prognosis assessment in intravenous thrombolytic therapy, segmentation model which is based on the self-attention mechanism is prone to generate attention coefficient maps with incorrect regions of interest. Moreover, the stroke lesion is not clearly characterized, and lesion boundary is poorly differentiated from normal brain tissue, thus affecting the segmentation performance. To address this problem, a primary and secondary path attention compensation network structure is proposed, which is based on the improved global attention upsampling U-Net model. The main path network is responsible for performing accurate lesion segmentation and outputting segmentation results. Likewise, the auxiliary path network generates loose auxiliary attention compensation coefficients, which compensate for possible attention coefficient errors in the main path network. Two hybrid loss functions are proposed to realize the respective functions of main and auxiliary path networks. It is experimentally demonstrated that both the improved global attention upsampling U-Net and the proposed primary and secondary path attention compensation networks show significant improvement in segmentation performance. Moreover, patients with good collateral circulation have a small final infarct area volume and a good clinical prognosis after intravenous thrombolysis. Quantitative assessment of collateral circulation and ischemic semidark zone by multimodal CT can better predict the clinical prognosis of intravenous thrombolysis.

Infarct Progression in the Early and Late Phases of Acute Ischemic Stroke

Purpose of ReviewTo explore factors associated with infarct progression in the early and late phase of acute ischemic stroke in patients undergoing endovascular therapy.Recent FindingsFollowing ischemic stroke, brain injury can progress at a variable rate, at the expense of “penumbral tissue,” which is the ischemic tissue at risk of infarction. Despite dramatic advances in endovascular stroke therapies with early revascularization in more than 80% of cases, nearly half of patients do not achieve functional independence despite successful recanalization. This is largely attributed to the irreversible damage that is already extensive at the time of revascularization.SummaryThe underlying pathophysiology and determinants of the core infarct progression are complex and multifactorial, depending on a balance between brain energy consumption and collateral perfusion supply. It is crucial to develop creative and individualized theranostics to predict infarct progression and to “freeze” the tissue at risk prior to recanalization.

Computed Tomography Perfusion–Based Prediction of Core Infarct and Tissue at Risk: Can Artificial Intelligence Help Reduce Radiation Exposure?

Background and Purpose: We explored the feasibility of automated, arterial input function independent, vendor neutral prediction of core infarct, and penumbral tissue using complete and partial computed tomographic perfusion data sets through neural networks. Methods: Using retrospective computed tomographic perfusion data from 57 patients, split as training/validation (60%/40%), we developed and validated separate 2-dimensional U-net models for cerebral blood flow (CBF) and time to maximum (Tmax) maps calculation to predict core infarct and tissue at risk, respectively. Once trained, the full sets of 28 input images were sequentially reduced to equitemporal 14, 10, and 7 time points. The averaged structural similarity index measure between the model-derived images and ground truth perfusion maps was compared. Volumes for core infarct and Tmax were compared using the Pearson correlation coefficient. Results: Both CBF and Tmax maps derived using 28 and 14 time points had similar structural similarity index measure (0.80–0.81; P >0.05) when compared with ground truth images. The Pearson correlation for the CBF and Tmax volumes derived from the model using 28-tp with ground truth volumes derived from the RAPID software was 0.69 for CBF and 0.74 for Tmax. The predicted maps were fully concordant in terms of laterality to the commercial perfusion maps. The mean Dice scores were 0.54 for the core infarct and 0.63 for the hypoperfusion maps. Conclusion: Artificial intelligence model-derived volumes show good correlation with RAPID-derived volumes for CBF and Tmax. Within the constraints of a small sample size, the perfusion map quality is similar when using 14-tp instead of 28-tp. Our findings provide proof of concept that vendor neutral artificial intelligence models for computed tomographic perfusion processing using complete or partial image data sets appear feasible. The model accuracy could be further optimized using larger data sets.

Abstract P500: Outcomes of Mechanical Thrombectomy in Patients With Low Aspects: Insights From Star

Introduction: Patients with poor baseline images were excluded from most clinical trials so the data about whether these patients could benefit from MT remains unknown. In this study, we aim to investigate the safety and efficacy of MT in patients with large vessel occlusion (LVO) and large core infarct (LCI). Methods: The Stroke Thrombectomy and Aneurysm Registry (STAR) was interrogated. We included thrombectomy patients presenting with LVO within 24 hours and with a LCI as defined by Alberta Stroke Program Early CT Score (ASPECTS) < 6. Patients presenting within 6 hours of last known normal (LKN) were considered in the early window and patients presenting after 6 hours were considered in the late window. 90-day outcomes were assessed. We used a logistic regression model to assess the factors associated with good 90-day outcome in patients in the early and late windows. Results: 144 patients were included in this study (table). Median age was 69 and 92 (64%) patients were treated in the early MT window. ICA was the most common site of occlusion (48.6%) and ADAPT was used in 34.7%. Admission NIHSS was 17.5. Successful recanalization (TICI>2b) was achieved in 84.7% and median procedure time was 54 minutes. sICH hemorrhage was observed in 22 (15.3%). Median mRS was 4 at 90 days. Favorable outcome was observed in 41 patients (28.5%) and mortality occurred in in 59 (41%). There was no difference in 90-day functional outcome between patients in early and late windows. In patients presenting in the early window, age (aOR=0.905, p=0.0002) and baseline NIHSS (aOR=0.909, p=0.0423) were independently associated with 90-day outcome. In patients presenting in the late window, only age (aOR=0.934, p=0.0069) was independently associated with good outcome. Conclusion: More than one in four patients presenting with ASPECTS<6 may achieve functional independence at 90-day following MT. Patient age remains the main predictor of 90-day outcome in patients with low ASPECTS in both late and early windows.

Abstract P468: Functional Outcome of Patients Undergoing Mechanical Thrombectomy Outside of the Dawn and Defuse Criteria

Background: The DAWN and DEFUSE trials (D&D) extended the mechanical thrombectomy (MT) window to 16-24 hours from last known normal (LKN) in patients with internal carotid artery (ICA) or proximal middle cerebral artery (M1) large vessel occlusions (LVO), with limits on core infarct and penumbra size. With advances in imaging and procedural techniques, many have extended these results to patients outside of the study criteria, yet there is limited analysis of outcomes in those receiving MT beyond these guidelines. This study evaluates the functional outcome of MT in acute stroke patients presenting within 24 hours of LKN outside of the D&D criteria vs those meeting study criteria at a single Comprehensive Stroke Center. Methods: An IRB approved study was performed in which consecutive anterior circulation ischemic stroke patients presenting to Wake Forest Baptist Health from 12/1/18-12/31/19 within 0-24 hours of LKN were retrospectively studied. Patients were grouped by eligibility for DAWN or DEFUSE (mRS ≦1, ICA/M1 occlusion, age <80 NIHSS ≧10 and core 0-30, or age <80 NIHSS ≧20 and core 31-51, or age ≧80 NIHSS ≧10 and core 0-20; or mRS ≦2, NIHSS ≧6, ICA/M1 occlusion with core <70, mismatch volume >15, and mismatch ratio >1.8) or neither. Good outcome was defined as mRS ≦3 within 3 months of discharge. If not available, discharge mRS was used. A multivariate logistic regression model analyzed the outcome controlling for confounding factors. Results: Of 130 patients, 57 (44%) fell outside D&D criteria. There was no significant difference between age, rate of IV tPA, admission NIHSS, baseline co-morbidities, or core infarct size between groups. Median mismatch volume was greater in the D&D group compared to patients in the non D&D group (97cc, IQR 67-146 vs 67cc, IQR 30-111; p=0.002). Good outcome was seen in 34 (47%) D&D patients vs 30 (53%) in the non D&D group; controlling for confounding factors affecting 3 month mRS (age, NIHSS, time to revascularization, atrial fibrillation, prior stroke), this was not statistically significant (p=0.25). Conclusion: The outcome of MT in patients with anterior circulation LVO is not different between those meeting and not meeting D&D criteria. MT should be considered in patients presenting within 24 hours outside study guidelines.

Effect of Oxygen Extraction (Brush-Sign) on Baseline Core Infarct Depends on Collaterals (HIR)

Objectives: Baseline-core-infarct volume is a critical factor in patient selection and outcome in acute ischemic stroke (AIS) before mechanical thrombectomy (MT). We determined whether oxygen extraction efficiency and arterial collaterals, two different physiologic components of the cerebral ischemic cascade, interacted to modulate baseline-core-infarct volume in patients with AIS-LVO undergoing MT triage.Methods: Between January 2015 and March 2018, consecutive patients with an AIS and M1 occlusion considered for MT with a baseline MRI and perfusion-imaging were included. Variables such as baseline-core-infarct volume [mL], arterial collaterals (HIR: TMax &gt; 10 s volume/TMax &gt; 6 s), high oxygen extraction (HOE, presence of the brush-sign on T2*) were assessed. A linear-regression was used to test the interaction of HOE and HIR with baseline-core-infarct volume, after including potential confounding variables.Results: We included 103 patients. Median age was 70 (58–78), and 63% were female. Median baseline-core-infarct volume was 32 ml (IQR 8–74.5). Seventy six patients (74%) had HOE. In a multivariate analysis both favorable HIR collaterals (p = 0.02) and HOE (p = 0.038) were associated with lower baseline-core-infarct volume. However, HOE significantly interacted with HIR (p = 0.01) to predict baseline-core-infarct volume, favorable collaterals (low HIR) with HOE was associated with small baseline-core-infarct whereas patients with poor collaterals (high HIR) and HOE had large baseline-core-infarct.Conclusion: While HOE under effective collateral blood-flow has the lowest baseline-core-infarct volume of all patients, the protective effect of HOE reverses under poor collateral blood-flow and may be a maladaptive response to ischemic stroke as measured by core infarctions in AIS-LVO patients undergoing MT triage.