Automatic Detection, Localization, and Volume Estimation of Ischemic Infarcts in Noncontrast Computed Tomographic Scans: Method and Preliminary Results

Haemorrhage after thrombectomy with adjuvant thrombolysis in unknown onset stroke depends on high early lesion water uptake

BACKGROUND AND PURPOSE

In wake-up stroke, CT-based quantitative net water uptake (NWU) might serve as an alternative tool to MRI to guide intravenous thrombolysis with alteplase (IVT). An important complication after IVT is symptomatic intracerebral haemorrhage (sICH). As NWU directly implies ischaemic lesion progression, reflecting blood-brain barrier injury, we hypothesised that NWU predicts sICH in patients who had a ischaemic stroke undergoing thrombectomy with unknown onset.

METHODS

Consecutive analysis of all patients who had unknown onset anterior circulation ischaemic stroke who underwent CT at baseline and endovascular treatment between December 2016 and October 2020. Quantitative NWU was assessed on baseline CT. The primary endpoint was sICH. The association of NWU and other baseline parameters to sICH was investigated using inverse-probability weighting (IPW) analysis.

RESULTS

A total of 88 patients were included, of which 46 patients (52.3%) received IVT. The median NWU was 10.7% (IQR: 5.1-17.7). The proportion of patients with any haemorrhage and sICH were 35.2% and 13.6%. NWU at baseline was significantly higher in patients with sICH (19.1% vs 9.6%, p<0.0001) and the median Alberta Stroke Program Early CT Score (ASPECTS) was lower (5 vs 8, p<0.0001). Following IPW, there was no association between IVT and sICH in unadjusted analysis. However, after adjusting for ASPECTS and NWU, there was a significant association between IVT administration and sICH (14.6%, 95% CI: 3.3% to 25.6%, p<0.01).

CONCLUSION

In patients with ischaemic stroke with unknown onset, the combination of high NWU with IVT is directly linked to higher rates of sICH. Besides ASPECTS for evaluating the extent of the early infarct lesion, quantitative NWU could be used as an imaging biomarker to assess the degree of blood-brain barrier damage in order to predict the risk of sICH in patients with wake up stroke.

Ischemic core detection threshold of computed tomography perfusion (CTP) in acute stroke

PURPOSE

This study aimed to determine the accuracy of detecting ischemic core volume using computed tomography perfusion (CTP) in patients with suspected acute ischemic stroke compared to diffusion-weighted magnetic resonance imaging (DW-MRI) as the reference standard.

METHODS

This retrospective monocentric study included patients who underwent CTP and DW-MRI for suspected acute ischemic stroke. The ischemic core size was measured at DW-MRI. The detectability threshold volume was defined as the lowest volume detected by each method. Clinical data on revascularization therapy, along with the clinical decision that influenced the choice, were collected. Volumes of the ischemic cores were compared using the Mann-Whitney U test.

RESULTS

Of 83 patients who underwent CTP, 52 patients (median age 73 years, IQR 63-80, 36 men) also had DW-MRI and were included, with a total of 70 ischemic cores. Regarding ischemic cores, only 18/70 (26%) were detected by both CTP and DW-MRI, while 52/70 (74%) were detected only by DW-MRI. The median volume of the 52 ischemic cores undetected on CTP (0.6 mL, IQR 0.2-1.3 mL) was significantly lower (p < 0.001) than that of the 18 ischemic cores detected on CTP (14.2 mL, IQR 7.0-18.4 mL). The smallest ischemic core detected on CTP had a volume of 5.0 mL. Among the 20 patients with undetected ischemic core on CTP, only 10% (2/20) received thrombolysis treatment.

CONCLUSIONS

CTP maps failed in detecting ischemic cores smaller than 5 mL. DW-MRI remains essential for suspected small ischemic brain lesions to guide a correct treatment decision-making.

Taxonomy of Acute Stroke: Imaging, Processing, and Treatment

Stroke management employs a variety of diagnostic imaging modalities, image processing and analysis methods, and treatment procedures. This work categorizes methods for stroke imaging, image processing and analysis, and treatment, and provides their taxonomies illustrated by a state-of-the-art review. Imaging plays a critical role in stroke management, and the most frequently employed modalities are computed tomography (CT) and magnetic resonance (MR). CT includes unenhanced non-contrast CT as the first-line diagnosis, CT angiography, and CT perfusion. MR is the most complete method to examine stroke patients. MR angiography is useful to evaluate the severity of artery stenosis, vascular occlusion, and collateral flow. Diffusion-weighted imaging is the gold standard for evaluating ischemia. MR perfusion-weighted imaging assesses the penumbra. The stroke image processing methods are divided into non-atlas/template-based and atlas/template-based. The non-atlas/template-based methods are subdivided into intensity and contrast transformations, local segmentation-related, anatomy-guided, global density-guided, and artificial intelligence/deep learning-based. The atlas/template-based methods are subdivided into intensity templates and atlases with three atlas types: anatomy atlases, vascular atlases, and lesion-derived atlases. The treatment procedures for arterial and venous strokes include intravenous and intraarterial thrombolysis and mechanical thrombectomy. This work captures the state-of-the-art in stroke management summarized in the form of comprehensive and straightforward taxonomy diagrams. All three introduced taxonomies in diagnostic imaging, image processing and analysis, and treatment are widely illustrated and compared against other state-of-the-art classifications.

Precision and Speed at Your Fingertips: An Automated Intracranial Hematoma Volume Calculation

BACKGROUND

Intracranial hemorrhage (ICH) is a severe condition that requires rapid diagnosis and treatment. Automated methods for calculating ICH volumes can reduce human error and improve clinical decisioPlease provide professional degrees (e.g., PhD, MD) for the corresponding author.n-making. A novel automated method has been developed that is comparable to the ABC/2 method in terms of speed and accuracy while providing more accurate volumetric data.

METHODS

We developed a novel automated algorithm for calculating intracranial blood volume from computed tomography (CT) scans. The algorithm consists of a Python script that processes Digital Imaging and Communications in Medicine images and determines the blood volume and ratio. The algorithm was validated against manual calculations performed by neurosurgeons.

RESULTS

Our novel automated algorithm for calculating intracranial blood volume from CT scans demonstrated excellent agreement with the ABC/2 method, with a median overall difference of just 1.46 mL. The algorithm was also validated in patient groups with ICH, epidural hematoma (EDH), and SDH, with agreement coefficients of 0.992, 0.983, and 0.997, respectively.

CONCLUSIONS

The study introduces a novel automated algorithm for calculating the volumes of various ICHs (EDH, and SDH) within CT scans. The algorithm showed excellent agreement with manual calculations and outperformed the commonly used ABC/2 method, which tends to overestimate ICH volume. The automated algorithm offers a more accurate, efficient, and time-saving approach to quantifying ICH, EDH, and SDH volumes, making it a valuable tool for clinical evaluation and decision-making.

Novel artificial intelligence-based hypodensity detection tool improves clinician identification of hypodensity on non-contrast computed tomography in stroke patients

Introduction

In acute stroke, identifying early changes (parenchymal hypodensity) on non-contrast CT (NCCT) can be challenging. We aimed to identify whether the accuracy of clinicians in detecting acute hypodensity in ischaemic stroke patients on a non-contrast CT is improved with the use of an Artificial Intelligence (AI) based, automated hypodensity detection algorithm (HDT) using MRI-DWI as the gold standard.

Methods

The study employed a case-crossover within-clinician design, where 32 clinicians were tasked with identifying hypodensity lesions on NCCT scans for five a priori selected patient cases, before and after viewing the AI-based HDT. The DICE similarity coefficient (DICE score) was the primary measure of accuracy. Statistical analysis compared DICE scores with and without AI-based HDT using mixed-effects linear regression, with individual NCCT scans and clinicians as nested random effects.

Results

The AI-based HDT had a mean DICE score of 0.62 for detecting hypodensity across all NCCT scans. Clinicians' overall mean DICE score was 0.33 (SD 0.31) before AI-based HDT implementation and 0.40 (SD 0.27) after implementation. AI-based HDT use was associated with an increase of 0.07 (95% CI: 0.02-0.11, p = 0.003) in DICE score accounting for individual scan and clinician effects. For scans with small lesions, clinicians achieved a mean increase in DICE score of 0.08 (95% CI: 0.02, 0.13, p = 0.004) following AI-based HDT use. In a subgroup of 15 trainees, DICE score improved with AI-based HDT implementation [mean difference in DICE 0.09 (95% CI: 0.03, 0.14, p = 0.004)].

Discussion

AI-based automated hypodensity detection has potential to enhance clinician accuracy of detecting hypodensity in acute stroke diagnosis, especially for smaller lesions, and notably for less experienced clinicians.

Localization of early infarction on non-contrast CT images in acute ischemic stroke with deep learning approach

Localization of early infarction on first-line Non-contrast computed tomogram (NCCT) guides prompt treatment to improve stroke outcome. Our previous study has shown a good performance in the identification of ischemic injury on NCCT. In the present study, we developed a deep learning (DL) localization model to help localize the early infarction sign on NCCT. This retrospective study included consecutive 517 ischemic stroke (IS) patients who received NCCT within 12 h after stroke onset. A total of 21,436 infarction patches and 20,391 non-infarction patches were extracted from the slice pool of 1,634 NCCT according to brain symmetricity property. The generated patches were fed into different pretrained convolutional neural network (CNN) models such as Visual Geometry Group 16 (VGG16), GoogleNet, Residual Networks 50 (ResNet50), Inception-ResNet-v2 (IR-v2), Inception-v3 and Inception-v4. The selected VGG16 model could detect the early infarction in both supratentorial and infratentorial regions to achieve an average area under curve (AUC) 0.73 after extensive customization. The properly tuned-VGG16 model could identify the early infarction in the cortical, subcortical and cortical plus subcortical areas of supratentorial region with the mean AUC > 0.70. Further, the model could attain 95.6% of accuracy on recognizing infarction lesion in 494 out of 517 IS patients.

Brain injury mobile diagnostic system: Applications in civilian medical service and on the battlefield-General concept and medical aspects

To present the concept of a portable ultrasound tomography device for diagnosing traumatic and vascular brain lesions. The device consisting of multiple transcranial ultrasound probes placed on the surface of the head, specifically but not exclusively in natural acoustic windows. An integral part of the mobile diagnostic system (MDS) is a decision support system based on artificial intelligence algorithms utilizing information from: head images, laboratory data, and assessment of the patient's clinical condition. The MDS can significantly reduce the time from stroke onset to rtPA therapy in civilian medical services and support therapeutic and evacuation strategies in instances of brain and skull trauma on the battlefield.

Association Between Net Water Uptake and Functional Outcome in Patients With Low ASPECTS Brain Lesions: Results From the I-LAST Study

BACKGROUND AND OBJECTIVES

The effect of mechanical thrombectomy (MT) on functional outcome in patients with ischemic stroke with low ASPECTS is still uncertain. ASPECTS rating is based on the presence of ischemic hypoattenuation relative to normal; however, the degree of hypoattenuation, which directly reflects net uptake of water, is currently not considered an imaging biomarker in stroke triage. We hypothesized that the effect of thrombectomy on functional outcome in low ASPECTS patients depends on early lesion water uptake.

METHODS

For this multicenter observational study, patients with anterior circulation stroke with ASPECTS ≤5 were consecutively analyzed. Net water uptake (NWU) was assessed as a quantitative imaging biomarker in admission CT. The primary end point was the rate of favorable functional outcome defined as modified Rankin Scale score 0-3 at day 90. The effect of recanalization on functional outcome was analyzed according to the degree of NWU within the early infarct lesion.

RESULTS

A total of 254 patients were included, of which 148 (58%) underwent MT. The median ASPECTS was 4 (interquartile range [IQR] 3-5), and the median NWU was 11.4% (IQR 8.9%-15.1%). The rate of favorable outcome was 27.6% in patients with low NWU (<11.4%) vs 6.3% in patients with high NWU (≥11.4%; p < 0.0001). In multivariable logistic regression analysis, NWU was an independent predictor of outcome, whereas vessel recanalization (modified thrombolysis in cerebral infarction ≥2b) was only significantly associated with better outcomes if NWU was lower than 12.6%. In inverse-probability weighting analysis, recanalization was associated with 20.7% (p = 0.01) increase in favorable outcome in patients with low NWU compared with 9.1% (p = 0.06) in patients with high NWU.

DISCUSSION

Early NWU was independently associated with clinical outcome and might serve as an indicator of futile MT in low ASPECTS patients. NWU could be tested as a tool to select low ASPECTS patients for MT.

TRIAL REGISTRATION INFORMATION

The study is registered within the ClinicalTrials.gov Protocol Registration and Results System (NCT04862507).

An Improved Detection Algorithm for Ischemic Stroke NCCT Based on YOLOv5

Cerebral stroke (CS) is a heterogeneous syndrome caused by multiple disease mechanisms. Ischemic stroke (IS) is a subtype of CS that causes a disruption of cerebral blood flow with subsequent tissue damage. Noncontrast computer tomography (NCCT) is one of the most important IS detection methods. It is difficult to select the features of IS CT within computational image analysis. In this paper, we propose AC-YOLOv5, which is an improved detection algorithm for IS. The algorithm amplifies the features of IS via an NCCT image based on adaptive local region contrast enhancement, which then detects the region of interest via YOLOv5, which is one of the best detection algorithms at present. The proposed algorithm was tested on two datasets, and seven control group experiments were added, including popular detection algorithms at present and other detection algorithms based on image enhancement. The experimental results show that the proposed algorithm has a high accuracy (94.1% and 91.7%) and recall (85.3% and 88.6%) rate; the recall result is especially notable. This proves the excellent performance of the accuracy, robustness, and generalizability of the algorithm.

The robust UCATR algorithm enhances the specificity and sensitivity to detect the infarct of acute ischaemic stroke within 6 hours of onset via non-contrast computed tomography images

PROBLEM BACKGROUND

Early detection of acute ischemic stroke (AIS) may provide patients with benefits against harmful health and financial impacts. The use of non-contrast computed tomography images for early detect of the infarct remains controversial.

MATERIALS & METHODS

Here, we used the UCATR algorithm to extract the pixel values of the infarct and the corresponding contralateral healthy area as the control surface in each NCCT slice for the whole brain. Magnetic resonance imaging results were used to verify both areas. We found significant pathological changes in the infarct compared with the corresponding contralateral healthy area in each NCCT slice.

ATTAINED RESULTS

Our approach validated that NCCT can be used to detect the lesion area in the early stage of AIS.

CONCLUSIONS

With obvious advantages such as saving time and the ability to quantify the infarct volume, this approach could help more patients survive the fatal and irreversible pathological process of AIS and improve their quality of life after AIS treatment.

Net water uptake, a neuroimaging marker of early brain edema, as a predictor of symptomatic intracranial hemorrhage after acute ischemic stroke

Objective

We hypothesized that quantitative net water uptake (NWU), a novel neuroimaging marker of early brain edema, can predict symptomatic intracranial hemorrhage (sICH) after acute ischemic stroke (AIS).

Methods

We enrolled patients with AIS who completed admission multimodal computed tomography (CT) within 24 h after stroke onset. NWU within the ischemic core and penumbra was calculated based on admission CT, namely NWU-core and NWU-penumbra. sICH was defined as the presence of ICH in the infarct area within 7 days after stroke onset, accompanied by clinical deterioration. The predictive value of NWU-core and NWU-penumbra on sICH was evaluated by logistic regression analyses and the receiver operating characteristic (ROC) curve. A pure neuroimaging prediction model was built considering imaging markers, which has the potential to be automatically quantified with an artificial algorithm on image workstation.

Results

154 patients were included, of which 93 underwent mechanical thrombectomy (MT). The median time from symptom onset to admission CT was 262 min (interquartile range, 198–368). In patients with MT, NWU-penumbra (OR =1.442; 95% CI = 1.177–1.766; P < 0.001) and NWU-core (OR = 1.155; 95% CI = 1.027–1.299; P = 0.016) were independently associated with sICH with adjustments for age, sex, time from symptom onset to CT, hypertension, lesion volume, and admission National Institutes of Health Stroke Scale (NIHSS) score. ROC curve showed that NWU-penumbra had better predictive performance than NWU-core on sICH [area under the curve (AUC): 0.773 vs. 0.673]. The diagnostic efficiency of the predictive model was improved with the containing of NWU-penumbra (AUC: 0.853 vs. 0.760). A pure imaging model also presented stable predictive power (AUC = 0.812). In patients without MT, however, only admission NIHSS score (OR = 1.440; 95% CI = 1.055–1.965; P = 0.022) showed significance in predicting sICH in multivariate analyses.

Conclusions

NWU-penumbra may have better predictive performance than NWU-core on sICH after MT. A pure imaging model showed potential value to automatically screen patients with sICH risk by image recognition, which may optimize treatment strategy.

Assessment of Irreversible Tissue Injury in Extensive Ischemic Stroke-Potential of Quantitative Cerebral Perfusion

Computed tomography perfusion (CTP) is used as a tool to select ischemic stroke patients for endovascular treatment (EVT) and is currently investigated in the setting of extensive stroke with low Alberta Stroke Program Early CT scores (ASPECTS). The purpose of this study was to perform a comprehensive quantitative analysis of cerebral blood flow within the ischemic lesion compared to threshold-derived core lesion volumes. We hypothesized that the degree of cerebral blood volume (CBV) reduction within the ischemic lesion is predictive of irreversible tissue injury and functional outcome in patients with low ASPECTS. Ischemic stroke patients with an ASPECTS ≤ 5 who received multimodal CT on admission and underwent thrombectomy were analyzed. The ischemic lesion on CTP was identified, and CTP-derived parameters were measured as absolute means within the lesion and relative to the physiological perfusion measured in a contralateral region of interest. The degree of irreversible tissue injury was assessed using quantitative net water uptake (NWU). Functional endpoint was good outcome defined as modified Rankin Scale (mRS) scores 0-3 at day 90. One hundred eleven patients were included. The median core lesion volume was 71 ml (IQR: 25-107), and the median quantitative NWU was 9.5% (IQR: 6-13). Relative CBV (rCBV) reduction and ASPECTS at baseline were independently associated with NWU in multivariable linear regression analysis (ß: 12.4, 95%CI: 6.0-18.9, p < 0.0001) and (ß: - 0.78, 95% CI: - 1.53 to - 0.02; p = 0.045), respectively. Furthermore, rCBV was significantly associated with good outcome in patients with core volumes > 50 ml (OR: 0.16, 95% CI: 0.05-0.49, p = 0.001). Our study shows that rCBV reduction serves as an early surrogate for increase of NWU as a marker of irreversible tissue injury and lesion progression. Thus, the analysis of rCBV reduction within ischemic lesions may add another dimension to acute stroke triage in addition to core volumes or ASPECTS as indicators of the infarct extent and viability.

New imaging score for outcome prediction in basilar artery occlusion stroke

OBJECTIVE

In ischemic posterior circulation stroke, the utilization of standardized image scores is not established in daily clinical practice. We aimed to test a novel imaging score that combines the collateral status with the rating of the posterior circulation Acute Stroke Prognosis Early CT score (pcASPECTS). We hypothesized that this score (pcASCO) predicts functional outcome and malignant cerebellar edema (MCE).

METHODS

Ischemic stroke patients with acute BAO who received multimodal-CT and underwent thrombectomy on admission at two comprehensive stroke centers were analyzed. The posterior circulation collateral score by van der Hoeven et al was added to the pcASPECTS to define pcASCO as a 20-point score. Multivariable logistic regression analyses were performed to predict functional independence at day 90, assessed using modified Rankin Scale scores, and occurrence of MCE in follow-up CT using the established Jauss scale score as endpoints.

RESULTS

A total of 118 patients were included, of which 84 (71%) underwent successful thrombectomy. Based on receiver operating characteristic curve analysis, pcASCO ≥ 14 classified functional independence with higher discriminative power (AUC: 0.83, 95%CI: 0.71-0.91) than pcASPECTS (AUC: 0.74). In multivariable logistic regression analysis, pcASCO was significantly and independently associated with functional independence (aOR: 1.91, 95%CI: 1.25-2.92, p = 0.003), and MCE (aOR: 0.71, 95%CI: 0.53-0.95, p = 0.02).

CONCLUSION

The pcASCO could serve as a simple and feasible imaging tool to assess BAO stroke patients on admission and might be tested as a complementary tool to select patients for thrombectomy in uncertain situations, or to predict clinical outcome.

KEY POINTS

• The neurological assessment of basilar artery occlusion stroke patients can be challenging and there are yet no validated imaging scores established in daily clinical practice. • The pcASCO combines the rating of early ischemic changes with the status of the intracranial posterior circulation collaterals. • The pcASCO showed high diagnostic accuracy to predict functional outcome and malignant cerebellar edema and could serve as a simple and feasible imaging tool to support treatment selection in uncertain situations, or to predict clinical outcome.

A Review on Computer Aided Diagnosis of Acute Brain Stroke

Amongst the most common causes of death globally, stroke is one of top three affecting over 100 million people worldwide annually. There are two classes of stroke, namely ischemic stroke (due to impairment of blood supply, accounting for ~70% of all strokes) and hemorrhagic stroke (due to bleeding), both of which can result, if untreated, in permanently damaged brain tissue. The discovery that the affected brain tissue (i.e., 'ischemic penumbra') can be salvaged from permanent damage and the bourgeoning growth in computer aided diagnosis has led to major advances in stroke management. Abiding to the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines, we have surveyed a total of 177 research papers published between 2010 and 2021 to highlight the current status and challenges faced by computer aided diagnosis (CAD), machine learning (ML) and deep learning (DL) based techniques for CT and MRI as prime modalities for stroke detection and lesion region segmentation. This work concludes by showcasing the current requirement of this domain, the preferred modality, and prospective research areas.

Quantitative Lesion Water Uptake as Stroke Imaging Biomarker: A Tool for Treatment Selection in the Extended Time Window?

BACKGROUND AND PURPOSE

Patients presenting in the extended time window may benefit from mechanical thrombectomy. However, selection for mechanical thrombectomy in this patient group has only been performed using specialized image processing platforms, which are not widely available. We hypothesized that quantitative lesion water uptake calculated in acute stroke computed tomography (CT) may serve as imaging biomarker to estimate ischemic lesion progression and predict clinical outcome in patients undergoing mechanical thrombectomy in the extended time window.

METHODS

All patients with ischemic anterior circulation stroke presenting within 4.5 to 24 hours after symptom onset who received initial multimodal CT between August 2014 and March 2020 and underwent mechanical thrombectomy were analyzed. Quantitative lesion net water uptake was calculated from the admission CT. Prediction of clinical outcome was assessed using univariable receiver operating characteristic curve analysis and logistic regression analyses.

RESULTS

One hundred two patients met the inclusion criteria. In the multivariable logistic regression analysis, net water uptake (odds ratio, 0.78 [95% CI, 0.64-0.95], P=0.01), age (odds ratio, 0.94 [95% CI, 0.88-0.99]; P=0.02), and National Institutes of Health Stroke Scale (odds ratio, 0.88 [95% CI, 0.79-0.99], P=0.03) were significantly and independently associated with favorable outcome (modified Rankin Scale score ≤1), adjusted for degree of recanalization and Alberta Stroke Program Early CT Score. A multivariable predictive model including the above parameters yielded the highest diagnostic ability in the classification of functional outcome, with an area under the curve of 0.88 (sensitivity 92.3%, specificity 82.9%).

CONCLUSIONS

The implementation of quantitative lesion water uptake as imaging biomarker in the diagnosis of patients with ischemic stroke presenting in the extended time window might improve clinical prognosis. Future studies could test this biomarker as complementary or even alternative tool to CT perfusion.

Blood-Based Biomarkers: A Forgotten Friend of Hyperacute Ischemic Stroke

Ischemic stroke (IS) is the second leading cause of death worldwide. Multimodal neuroimaging techniques that have significantly facilitated the diagnosis of hyperacute IS are not widely used in underdeveloped areas and community hospitals owing to drawbacks such as high cost and lack of trained operators. Moreover, these methods do not have sufficient resolution to detect changes in the brain at the cellular and molecular levels after IS onset. In contrast, blood-based biomarkers can reflect molecular and biochemical alterations in both normal and pathophysiologic processes including angiogenesis, metabolism, inflammation, oxidative stress, coagulation, thrombosis, glial activation, and neuronal and vascular injury, and can thus provide information complementary to findings from routine examinations and neuroimaging that is useful for diagnosis. In this review, we summarize the current state of knowledge on blood-based biomarkers of hyperacute IS including those associated with neuronal injury, glial activation, inflammation and oxidative stress, vascular injury and angiogenesis, coagulation and thrombosis, and metabolism as well as genetic and genomic biomarkers. Meanwhile, the blood sampling time of the biomarkers which are cited and summarized in the review is within 6 h after the onset of IS. Additionally, we also discuss the diagnostic and prognostic value of blood-based biomarkers in stroke patients, and future directions for their clinical application and development.

Rapid Assessment of Acute Ischemic Stroke by Computed Tomography Using Deep Convolutional Neural Networks

Acute stroke is one of the leading causes of disability and death worldwide. Regarding clinical diagnoses, a rapid and accurate procedure is necessary for patients suffering from acute stroke. This study proposes an automatic identification scheme for acute ischemic stroke using deep convolutional neural networks (DCNNs) based on non-contrast computed tomographic (NCCT) images. Our image database for the classification model was composed of 1254 grayscale NCCT images from 96 patients (573 images) with acute ischemic stroke and 121 normal controls (681 images). According to the consensus of critical stroke findings by two neuroradiologists, a gold standard was established and used to train the proposed DCNN using machine-generated image features. Including the earliest DCNN, AlexNet, the popular Inception-v3, and ResNet-101 were proposed. To train the limited data size, transfer learning with ImageNet parameters was also used. The established models were evaluated by tenfold cross-validation and tested on an independent dataset containing 50 patients with acute ischemic stroke (108 images) and 58 normal controls (117 images) from another institution. AlexNet without pretrained parameters achieved an accuracy of 97.12%, a sensitivity of 98.11%, a specificity of 96.08%, and an area under the receiver operating characteristic curve (AUC) of 0.9927. Using transfer learning, transferred AlexNet, transferred Inception-v3, and transferred ResNet-101 achieved accuracies between 90.49 and 95.49%. Tested with a dataset from another institution, AlexNet showed an accuracy of 60.89%, a sensitivity of 18.52%, and a specificity of 100%. Transferred AlexNet, Inception-v3, and ResNet-101 achieved accuracies of 81.77%, 85.78%, and 80.89%, respectively. The proposed DCNN architecture as a computer-aided diagnosis system showed that training from scratch can generate a customized model for a specific scanner, and transfer learning can generate a more generalized model to provide diagnostic suggestions of acute ischemic stroke to radiologists.

Early Prediction of Malignant Cerebellar Edema in Posterior Circulation Stroke Using Quantitative Lesion Water Uptake

BACKGROUND

Malignant cerebellar edema (MCE) is a life-threatening complication of ischemic posterior circulation stroke that requires timely diagnosis and management. Yet, there is no established imaging biomarker that may serve as predictor of MCE. Early edematous water uptake can be determined using quantitative lesion water uptake, but this biomarker has only been applied in anterior circulation strokes.

OBJECTIVE

To test the hypothesis that lesion water uptake in early posterior circulation stroke predicts MCE.

METHODS

A total 179 patients with posterior circulation stroke and multimodal admission CT were included. A total of 35 (19.5%) patients developed MCE defined by using an established 10-point scale in follow-up CT, of which ≥4 points are considered malignant. Posterior circulation net water uptake (pcNWU) was quantified in admission CT based on CT densitometry and compared with posterior circulation Acute Stroke Prognosis Early CT Score (pc-ASPECTS) as predictor of MCE using receiver operating curve (ROC) analysis and logistic regression analysis.

RESULTS

Acute pcNWU within the early ischemic lesion was 24.6% (±8.4) for malignant and 7.2% (±7.4) for nonmalignant infarctions, respectively (P < .0001). Based on ROC analysis, pcNWU above 14.9% identified MCE with high discriminative power (area under the curve: 0.94; 95% CI: 0.89-0.97). Early pcNWU (odds ratio [OR]: 1.28; 95% CI: 1.15-1.42, P < .0001) and pc-ASPECTS (OR: 0.71, 95% CI: 0.53-0.95, P = .02) were associated with MCE, adjusted for age and recanalization status.

CONCLUSION

Quantitative pcNWU in early posterior circulation stroke is an important marker for MCE. Besides pc-ASPECTS, lesion water uptake measurements may further support identifying patients at risk for MCE at an early stage indicating stricter monitoring and consideration for further therapeutic measures.

Detecting the Early Infarct Core on Non-Contrast CT Images with a Deep Learning Residual Network

PURPOSE

To explore a new approach mainly based on deep learning residual network (ResNet) to detect infarct cores on non-contrast CT images and improve the accuracy of acute ischemic stroke diagnosis.

METHODS

We continuously enrolled magnetic resonance diffusion weighted image (MR-DWI) confirmed first-episode ischemic stroke patients (onset time: less than 9 h) as well as some normal individuals in this study. They all underwent CT plain scan and MR-DWI scan with same scanning range, layer thickness (4 mm) and interlayer spacing (4 mm) (The time interval between two examinations: less than 4 h). Setting MR-DWI as gold standard of infarct core and using deep learning ResNet combined with a maximum a posteriori probability (MAP) model and a post-processing method to detect the infarct core on non-contrast CT images. After that, we use decision curve analysis (DCA) establishing models to analyze the value of this new method in clinical practice.

RESULTS

116 ischemic stroke patients and 26 normal people were enrolled. 58 patients were allocated into training dataset and 58 were divided into testing dataset along with 26 normal samples. The identification accuracy of our ResNet based approach in detecting the infarct core on non-contrast CT is 75.9%. The DCA shows that this deep learning method is capable of improving the net benefit of ischemic stroke patients.

CONCLUSIONS

Our deep learning residual network assisted with optimization methods is able to detect early infarct core on non-contrast CT images and has the potential to help physicians improve diagnostic accuracy in acute ischemic stroke patients.

Ischemic infarct detection, localization, and segmentation in noncontrast CT human brain scans: review of automated methods

Noncontrast Computed Tomography (NCCT) of the brain has been the first-line diagnosis for emergency evaluation of acute stroke, so a rapid and automated detection, localization, and/or segmentation of ischemic lesions is of great importance. We provide the state-of-the-art review of methods for automated detection, localization, and/or segmentation of ischemic lesions on NCCT in human brain scans along with their comparison, evaluation, and classification. Twenty-two methods are (1) reviewed and evaluated; (2) grouped into image processing and analysis-based methods (11 methods), brain atlas-based methods (two methods), intensity template-based methods (1 method), Stroke Imaging Marker-based methods (two methods), and Artificial Intelligence-based methods (six methods); and (3) properties of these groups of methods are characterized. A new method classification scheme is proposed as a 2 × 2 matrix with local versus global processing and analysis, and density versus spatial sampling. Future studies are necessary to develop more efficient methods directed toward deep learning methods as well as combining the global methods with a high sampling both in space and density for the merged radiologic and neurologic data.

Human Brain Atlases in Stroke Management

Stroke is a leading cause of death and a major cause of permanent disability. Its management is demanding because of variety of protocols, imaging modalities, pulse sequences, hemodynamic maps, criteria for treatment, and time constraints to promptly evaluate and treat. To cope with some of these issues, we propose novel, patented solutions in stroke management by employing multiple brain atlases for diagnosis, treatment, and prediction. Numerous and diverse CT and MRI scans are used: ARIC cohort, ischemic and hemorrhagic stroke CT cases, MRI cases with multiple pulse sequences, and 128 stroke CT patients, each with 170 variables and one year follow-up. The method employs brain atlases of anatomy, blood supply territories, and probabilistic stroke atlas. It rapidly maps an atlas to scan and provides atlas-assisted scan processing. Atlas-to-scan mapping is application-dependent and handles three types of regions of interest (ROIs): atlas-defined ROIs, atlas-quantified ROIs, and ROIs creating an atlas. An ROI is defined by atlas-guided anatomy or scan-derived pathology. The atlas defines ROI or quantifies it. A brain atlas potential has been illustrated in four atlas-assisted applications for stroke occurrence prediction and screening, rapid and automatic stroke diagnosis in emergency room, quantitative decision support in thrombolysis in ischemic stroke, and stroke outcome prediction and treatment assessment. The use of brain atlases in stroke has many potential advantages, including rapid processing, automated and robust handling, wide range of applications, and quantitative assessment. Further work is needed to enhance the developed prototypes, clinically validate proposed solutions, and introduce them to clinical practice.

Automated stroke lesion segmentation in non-contrast CT scans using dense multi-path contextual generative adversarial network

Stroke lesion volume is a key radiologic measurement in assessing prognosis of acute ischemic stroke (AIS) patients. The aim of this paper is to develop an automated segmentation method for accurately segmenting follow-up ischemic and hemorrhagic lesion from multislice non-contrast CT (NCCT) volumes of AIS patients. This paper proposes a 2D dense multi-path contextual generative adversarial network (MPC-GAN) where a dense multi-path 2D U-Net is utilized as the generator and a discriminator network is applied to regularize the generator. Contextual information (i.e. bilateral intensity difference, distance map and lesion location probability) are input into the generator and discriminator. The proposed method is validated separately on follow-up NCCT volumes of 60 patients with ischemic infarcts and NCCT volumes of 70 patients with hemorrhages. Quantitative results demonstrated that the proposed MPC-GAN method obtained a Dice coefficient (DC) of 70.6% for ischemic infarct segmentation and a DC of 76.5% for hemorrhage segmentation compared with manual segmented lesions, outperforming several benchmark methods. Additional volumetric analyses demonstrated that the MPC-GAN segmented lesion volume correlated well with manual measurements (Pearson correlation coefficients were 0.926 and 0.927 for ischemic infarcts and hemorrhages, respectively). The proposed MPC-GAN method can accurately segment ischemic infarcts and hemorrhages from NCCT volumes of AIS patients.

Lesion Age Imaging in Acute Stroke: Water Uptake in CT Versus DWI-FLAIR Mismatch

PURPOSE

In acute ischemic stroke with unknown time of onset, magnetic resonance (MR)-based diffusion-weighted imaging (DWI) and fluid-attenuated inversion recovery (FLAIR) estimates lesion age to guide intravenous thrombolysis. Computed tomography (CT)-based quantitative net water uptake (NWU) may be a potential alternative. The purpose of this study was to directly compare CT-based NWU to magnetic resonance imaging (MRI) at identifying patients with lesion age < 4.5 hours from symptom onset.

METHODS

Fifty patients with acute anterior circulation stroke were analyzed with both imaging modalities at admission between 0.5 and 8.0 hours after known symptom onset. DWI-FLAIR lesion mismatch was rated and NWU was measured in admission CT. An established NWU threshold (11.5%) was used to classify patients within and beyond 4.5 hours. Multiparametric MRI signal was compared with NWU using logistic regression analyses. The empirical distribution of NWU was analyzed in a consecutive cohort of patients with wake-up stroke.

RESULTS

The median time between CT and MRI was 35 minutes (interquartile range [IQR] = 24-50). The accuracy of DWI-FLAIR mismatch was 68.8% (95% confidence interval [CI] = 53.7-81.3%) with a sensitivity of 58% and specificity of 82%. The accuracy of NWU threshold was 86.0% (95% CI = 73.3-94.2%) with a sensitivity of 91% and specificity of 78%. The area under the curve (AUC) of multiparametric MRI signal to classify lesion age <4.5 hours was 0.86 (95% CI = 0.64-0.97), and the AUC of quantitative NWU was 0.91 (95% CI = 0.78-0.98). Among 87 patients with wake-up stroke, 46 patients (53%) showed low NWU (< 11.5%).

CONCLUSION

The predictive power of CT-based lesion water imaging to identify patients within the time window of thrombolysis was comparable to multiparametric DWI-FLAIR MRI. A significant proportion of patients with wake-up stroke exhibit low NWU and may therefore be potentially suitable for thrombolysis. ANN NEUROL 2020;88:1144-1152.

Impact of endovascular recanalization on quantitative lesion water uptake in ischemic anterior circulation strokes

Studies evaluating the effect of reperfusion on ischemic edema in acute stroke described conflicting results. Net water uptake (NWU) per brain volume is a new quantitative imaging biomarker of space-occupying ischemic edema, which can be measured in computed tomography (CT). We sought to investigate the effects of vessel recanalization on the formation of ischemic brain edema using quantitative NWU. In this multicenter observational study, acute ischemic stroke patients with a large vessel occlusion (LVO) in the anterior circulation were consecutively screened. Patients with vessel recanalization (thrombolysis in cerebral infarction (TICI) 2 b or 3) versus persistent vessel occlusion (no thrombectomy, TICI 0-1) were compared. Lesion-NWU was quantified in multimodal admission CT and follow-up CT (FCT), and ΔNWU was calculated as difference. Of 194 included patients, 150 had successful endovascular recanalization and 44 persistent LVO. In FCT after treatment, the mean (standard deviation) ΔNWU was 15.8% (5.7) in patients with persistent LVO and 9.8% (5.8) with vessel recanalization (p < 0.001). In multivariate regression analysis, vessel recanalization was independently associated with a lowered ΔNWU by 6.3% compared to LVO (95% confidence interval: 3.7-9.0, p < 0.001). Successful vessel recanalization was associated with a significantly reduced formation of ischemic brain edema. Quantitative NWU may be used to compare the treatment effects in acute stroke.

Machine Learning for Detecting Early Infarction in Acute Stroke with Non-Contrast-enhanced CT

Background Identifying the presence and extent of infarcted brain tissue at baseline plays a crucial role in the treatment of patients with acute ischemic stroke (AIS). Patients with extensive infarction are unlikely to benefit from thrombolysis or thrombectomy procedures. Purpose To develop an automated approach to detect and quantitate infarction by using non-contrast-enhanced CT scans in patients with AIS. Materials and Methods Non-contrast-enhanced CT images in patients with AIS (<6 hours from symptom onset to CT) who also underwent diffusion-weighted (DW) MRI within 1 hour after AIS were obtained from May 2004 to July 2009 and were included in this retrospective study. Ischemic lesions manually contoured on DW MRI scans were used as the reference standard. An automatic segmentation approach involving machine learning (ML) was developed to detect infarction. Randomly selected nonenhanced CT images from 157 patients with the lesion labels manually contoured on DW MRI scans were used to train and validate the ML model; the remaining 100 patients independent of the derivation cohort were used for testing. The ML algorithm was quantitatively compared with the reference standard (DW MRI) by using Bland-Altman plots and Pearson correlation. Results In 100 patients in the testing data set (median age, 69 years; interquartile range [IQR]: 59-76 years; 59 men), baseline non-contrast-enhanced CT was performed within a median time of 48 minutes from symptom onset (IQR, 27-93 minutes); baseline MRI was performed a median of 38 minutes (IQR, 24-48 minutes) later. The algorithm-detected lesion volume correlated with the reference standard of expert-contoured lesion volume in acute DW MRI scans (r = 0.76, P < .001). The mean difference between the algorithm-segmented volume (median, 15 mL; IQR, 9-38 mL) and the DW MRI volume (median, 19 mL; IQR, 5-43 mL) was 11 mL (P = .89). Conclusion A machine learning approach for segmentation of infarction on non-contrast-enhanced CT images in patients with acute ischemic stroke showed good agreement with stroke volume on diffusion-weighted MRI scans. © RSNA, 2020 Online supplemental material is available for this article. See also the editorial by Nael in this issue.

Futile Recanalization With Poor Clinical Outcome Is Associated With Increased Edema Volume After Ischemic Stroke

PURPOSE

Futile recanalization with poor clinical outcome after endovascular treatment of acute ischemic stroke is poorly understood. Recently, vessel recanalization has been associated with reduced ischemic brain edema in patients with good clinical outcome. As edema volume (EV) may be quantified in computed tomography (CT), we hypothesized that higher EV after revascularization predicts futile recanalization with poor outcome.

METHODS

In this observational study, 67 ischemic stroke patients with M1 middle cerebral artery occlusion fulfilled all inclusion criteria and were analyzed. All patients received successful endovascular recanalization (thrombolysis in cerebral infarction scale 2b/3) and subsequent follow-up CT 24 hours later. Edema volume within the infarct lesion was calculated in follow-up CT applying lesion water uptake quantification and was used to predict clinical outcome (Modified Rankin Scale [mRS] after 90 days) compared with infarct volume.

RESULTS

The median EV after thrombectomy was 1.6 mL (interquartile range, 0.2-4.2 mL) in patients with mRS 0 to 4 and 8.6 mL (interquartile range, 2.0-49.8 mL) in patients with mRS 5 to 6 (P = 0.0008). In regression analysis, an EV increase of 1 mL was associated with an 8.0% increased likelihood of poor outcome (95% confidence interval, 2.8%-15.4%; P = 0.008). Based on univariate receiver operating characteristic curve analysis, absolute EV over 4.2 mL predicted poor outcome (mRS 5-6) with good discriminative power (area under curve, 0.74; 95% confidence interval, 0.62-0.84; specificity, 77%; sensitivity, 68%). In comparison, the area under curve for infarct volume was 0.68.

CONCLUSIONS

Elevated EV after endovascular thrombectomy was associated with poor clinical outcome and may indicate futile recanalization.

Computed Tomography Angiography Collateral Profile Is Directly Linked to Early Edema Progression Rate in Acute Ischemic Stroke

Background and Purpose—

Poor collateral flow is associated with poor clinical outcome in acute ischemic stroke and may indicate futile recanalization after successful thrombectomy. Pronounced early formation of cerebral ischemic edema may be the link between poor collateral status and declined functional outcome, but this relationship has not been investigated yet. We hypothesized that collateral status is associated with early lesion water uptake as quantitative marker for edema progression.

Methods—

One hundred seventy-six patients with middle cerebral artery stroke who underwent mechanical thrombectomy were analyzed. Status of cerebral collateral circulation (collaterals status [CS]) was derived using an established 5-point scoring system in admission computed tomography angiography, and good collaterals were defined as CS 3 to 4. Ischemic brain edema dynamics were quantified using early edema progression rate (EPR). EPR was derived from quantitative lesion water uptake in admission computed tomography divided by time from symptom onset to imaging. Good clinical outcome was defined as modified Rankin Scale score 0 to 2 after 90 days.

Results—

The median EPR was 1.4% per hour (interquartile range, 0.5–3.5%) in patients with good collaterals, which was lower than the median EPR in patients with poor collaterals of 5.8% per hour (interquartile range, 2.1–5.9%; P <0.0001). In multivariable regression analysis, lower CS was significantly and independently associated with higher EPR (1.6% EPR per 1-point CS; P =0.002). A higher EPR was associated with reduced likelihood of good clinical outcome: odds ratio 0.87; (95% CI, 0.76–0.99; P =0.03).

Conclusions—

Patients with poor CS had significantly higher EPR, which was associated with worse clinical outcome. These patients might benefit from adjuvant antiedematous treatment.

Elevated early lesion water uptake in acute stroke predicts poor outcome despite successful recanalization – When “tissue clock” and “time clock” are desynchronized

Background

Ischemic water uptake in acute stroke is a reliable indicator of lesion age. Nevertheless, inter-individually varying edema progression has been observed and elevated water uptake has recently been described as predictor of malignant infarction.

Aims

We hypothesized that early-elevated lesion water uptake indicates accelerated “tissue clock” desynchronized with “time clock” and therefore predicts poor clinical outcome despite successful recanalization.

Methods

Acute middle cerebral artery stroke patients with multimodal admission-CT who received successful thrombectomy (TICI 2b/3) were analyzed. Net water uptake (NWU), a quantitative imaging biomarker of ischemic edema, was determined in admission-CT and tested as predictor of clinical outcome using modified Rankin Scale (mRS) after 90 days. A binary outcome was defined for mRS 0–4 and mRS 5–6.

Results

Seventy-two patients were included. The mean NWU (SD) in patients with mRS 0–4 was lower compared to patients with mRS 5–6 (5.0% vs. 12.1%; p < 0.001) with similar time from symptom onset to imaging (2.6 h vs. 2.4 h; p = 0.7). Based on receiver operating curve analysis, NWU above 10% identified patients with very poor outcome with high discriminative power (AUC 0.85), followed by Alberta Stroke Program Early CT Score (ASPECTS) (AUC: 0.72) and National Institutes of Health Stroke Scale (NIHSS) (AUC: 0.72).

Conclusions

Quantitative NWU may serve as an indicator of “tissue clock” and pronounced early brain edema with elevated NWU might suggest a desynchronized “tissue clock” with real “time clock” and therefore predict futile recanalization with poor clinical outcome.

Quantitative Lesion Water Uptake in Acute Stroke Computed Tomography Is a Predictor of Malignant Infarction

Background and Purpose- Early selection of patients with acute middle cerebral artery infarction at risk for malignant edema is critical to initiate timely decompressive surgery. Net water uptake (NWU) per brain volume is a quantitative imaging biomarker of space-occupying ischemic edema which can be measured in computed tomography. We hypothesize that NWU in early infarct lesions can predict development of malignant edema. The aim was to compare NWU in acute brain infarct against other common predictors of malignant edema. Methods- After consecutive screening of single-center registry data, 153 patients with acute proximal middle cerebral artery occlusion fulfilled the inclusion criteria. A total of 29 (18.2%) patients developed malignant edema defined as end point in follow-up imaging leading to decompressive surgery and death as a direct implication of mass effect. Early infarct lesion volume and NWU were quantified in multimodal admission computed tomography; time from symptom onset to admission imaging was recorded. Results- Mean time from onset to admission imaging was equivalent between patients with and without malignant infarcts (mean±SD: 3.3±1.4 hours and 3.3±1.7 hours, respectively). Edematous tissue expansion by NWU within infarct lesions occurred across all patients in this cohort (NWU: 9.1%±6.8%; median, 7.9%; interquartile range, 8.8%; range, 0.1%-35.6%); 7.0% (±5.2) in nonmalignant and 18.0% (±5.7) in malignant infarcts. Based on univariate receiver operating characteristic curve analysis, NWU >12.7% or an edema rate >3.7% NWU/h identified malignant infarcts with high discriminative power (area under curve, 0.93±0.02). In multivariate binary logistic regression, the probability of malignant infarct was significantly associated with early infarct volume and NWU. Conclusions- Computed tomography-based quantitative NWU in early infarct lesions is an important surrogate marker for developing malignant edema. Besides volume of early infarct, the measurements of lesion water uptake may further support identifying patients at risk for malignant infarction.

Computed Tomography-Based Imaging of Voxel-Wise Lesion Water Uptake in Ischemic Brain: Relationship Between Density and Direct Volumetry

OBJECTIVES

Net water uptake per volume of brain tissue may be calculated by computed tomography (CT) density, and this imaging biomarker has recently been investigated as a predictor of lesion age in acute stroke. However, the hypothesis that measurements of CT density may be used to quantify net water uptake per volume of infarct lesion has not been validated by direct volumetric measurements so far. The purpose of this study was to (1) develop a theoretical relationship between CT density reduction and net water uptake per volume of ischemic lesions and (2) confirm this relationship by quantitative in vitro and in vivo CT image analysis using direct volumetric measurements.

MATERIALS AND METHODS

We developed a theoretical rationale for a linear relationship between net water uptake per volume of ischemic lesions and CT attenuation. The derived relationship between water uptake and CT density was tested in vitro in a set of increasingly diluted iodine solutions with successive CT measurements. Furthermore, the consistency of this relationship was evaluated using human in vivo CT images in a retrospective multicentric cohort. In 50 edematous infarct lesions, net water uptake was determined by direct measurement of the volumetric difference between the ischemic and normal hemisphere and was correlated with net water uptake calculated by ischemic density measurements.

RESULTS

With regard to in vitro data, water uptake by density measurement was equivalent to direct volumetric measurement (r = 0.99, P < 0.0001; mean ± SD difference, -0.29% ± 0.39%, not different from 0, P < 0.0001). In the study cohort, the mean ± SD uptake of water within infarct measured by volumetry was 44.7 ± 26.8 mL and the mean percent water uptake per lesion volume was 22.7% ± 7.4%. This was equivalent to percent water uptake obtained from density measurements: 21.4% ± 6.4%. The mean difference between percent water uptake by direct volumetry and percent water uptake by CT density was -1.79% ± 3.40%, which was not significantly different from 0 (P < 0.0001).

CONCLUSIONS

Volume of water uptake in infarct lesions can be calculated quantitatively by relative CT density measurements. Voxel-wise imaging of water uptake depicts lesion pathophysiology and could serve as a quantitative imaging biomarker of acute infarct lesions.

3D convolutional neural networks applied to CT angiography in the detection of acute ischemic stroke

Background

The aim of this study was to investigate the feasibility of ischemic stroke detection from computed tomography angiography source images (CTA-SI) using three-dimensional convolutional neural networks.

Methods

CTA-SI of 60 patients with a suspected acute ischemic stroke of the middle cerebral artery were randomly selected for this study; 30 patients were used in the neural network training, and the subsequent testing was performed using the remaining 30 patients. The training and testing were based on manually segmented lesions. Cerebral hemispheric comparison CTA and non-contrast computed tomography (NCCT) were studied as additional input features.

Results

All ischemic lesions in the testing data were correctly lateralized, and a high correspondence to manual segmentations was achieved. Patients with a diagnosed stroke had clinically relevant regions labeled infarcted with a 0.93 sensitivity and 0.82 specificity. The highest achieved voxel-wise area under receiver operating characteristic curve was 0.93, and the highest Dice similarity coefficient was 0.61. When cerebral hemispheric comparison was used as an input feature, the algorithm performance improved. Only a slight effect was seen when NCCT was included.

Conclusion

The results support the hypothesis that an acute ischemic stroke lesion can be detected with 3D convolutional neural network-based software from CTA-SI. Utilizing information from the contralateral hemisphere appears to be beneficial for reducing false positive findings.

A quantitative symmetry-based analysis of hyperacute ischemic stroke lesions in noncontrast computed tomography

PURPOSE

Early identification of ischemic stroke plays a significant role in treatment and potential recovery of damaged brain tissue. In noncontrast CT (ncCT), the differences between ischemic changes and healthy tissue are usually very subtle during the hyperacute phase (< 8 h from the stroke onset). Therefore, visual comparison of both hemispheres is an important step in clinical assessment. A quantitative symmetry-based analysis of texture features of ischemic lesions in noncontrast CT images may provide an important information for differentiation of ischemic and healthy brain tissue in this phase.

METHODS

One hundred thirty-nine (139) ncCT scans of hyperacute ischemic stroke with follow-up magnetic resonance diffusion-weighted (MR-DW) images were collected. The regions of stroke were identified in the MR-DW images, which were spatially aligned to corresponding ncCT images. A state-of-the-art symmetric diffeomorphic image registration was utilized for the alignment of CT and MR-DW, for identification of individual brain hemispheres, and for localization of the region representing healthy tissue contralateral to the stroke cores. Texture analysis included extraction and classification of co-occurrence and run-length texture-based image features in the regions of ischemic stroke and their contralateral regions.

RESULTS

The classification schemes achieved area under the receiver operating characteristic [Az] ≈ 0.82 for the whole dataset. There was no statistically significant difference in the performance of classifiers for the data sets with time between 2 and 8 hours from symptom onset. The performance of the classifiers did not depend on the size of the stroke regions.

CONCLUSIONS

The results provide a set of optimal texture features which are suitable for distinguishing between hyperacute ischemic lesions and their corresponding contralateral brain tissue in noncontrast CT. This work is an initial step toward development of an automated decision support system for detection of hyperacute ischemic stroke lesions on noncontrast CT of the brain.

Development and Validation of an Automatic Segmentation Algorithm for Quantification of Intracerebral Hemorrhage

BACKGROUND AND PURPOSE

ABC/2 is still widely accepted for volume estimations in spontaneous intracerebral hemorrhage (ICH) despite known limitations, which potentially accounts for controversial outcome-study results. The aim of this study was to establish and validate an automatic segmentation algorithm, allowing for quick and accurate quantification of ICH.

METHODS

A segmentation algorithm implementing first- and second-order statistics, texture, and threshold features was trained on manual segmentations with a random-forest methodology. Quantitative data of the algorithm, manual segmentations, and ABC/2 were evaluated for agreement in a study sample (n=28) and validated in an independent sample not used for algorithm training (n=30).

RESULTS

ABC/2 volumes were significantly larger compared with either manual or algorithm values, whereas no significant differences were found between the latter (P<0.0001; Friedman+Dunn's multiple comparison). Algorithm agreement with the manual reference was strong (concordance correlation coefficient 0.95 [lower 95% confidence interval 0.91]) and superior to ABC/2 (concordance correlation coefficient 0.77 [95% confidence interval 0.64]). Validation confirmed agreement in an independent sample (algorithm concordance correlation coefficient 0.99 [95% confidence interval 0.98], ABC/2 concordance correlation coefficient 0.82 [95% confidence interval 0.72]). The algorithm was closer to respective manual segmentations than ABC/2 in 52/58 cases (89.7%).

CONCLUSIONS

An automatic segmentation algorithm for volumetric analysis of spontaneous ICH was developed and validated in this study. Algorithm measurements showed strong agreement with manual segmentations, whereas ABC/2 exhibited its limitations, yielding inaccurate overestimations of ICH volume. The refined, yet time-efficient, quantification of ICH by the algorithm may facilitate evaluation of clot volume as an outcome predictor and trigger for surgical interventions in the clinical setting.

A Quantitative Method Using Head Noncontrast CT Scans to Detect Hyperacute Nonvisible Ischemic Changes in Patients With Stroke

PURPOSE

Because clinical evaluation of noncontrast computed tomography (CT) has a poor sensitivity in the evaluation of acute ischemic stroke, computer-aided diagnosis may be able to facilitate the performance. Recently, we introduced a computational method for the detection and localization of visible infarcts. Herein, we aimed to evaluate and extend a previous method, the Stroke Imaging Marker (SIM), to localize nonvisible hyperacute ischemia.

MATERIALS AND METHODS

On the basis of the SIM and its components-the ratio of percentile differences in subranges of Hounsfield Unit (HU) distribution (P-ratio), ratio of voxels count in ranges of brain CT intensity, median HU attenuation value-the infarct localization was performed in 140 early and follow-up scans of 70 patients. In none of the early scans was the infarct visible to a radiologist or an experienced stroke neuroradiologist. The infarcted hemisphere detection rate (HDR) and sensitivity of infarct localization were measured by overlapping the region of detected tissue in the initial scan, with the gold standard set for the fully visible stroke in the follow-up scan.

RESULTS

The best performance of the algorithm was found for the P-ratio including seven percentile subranges within the range of 35th-75th percentile. The modified SIM provided a 76% ischemic HDR and 54% sensitivity in spatial localization of hyperacute ischemia (68% among properly detected infarct sides).

CONCLUSION

The improved SIM is a dedicated and potentially useful tool for hyperacute nonvisible brain infarct detection from CT scans and may contribute to reduction of image-to-needle time in patients eligible for revascularization therapy.

Diagnostic accuracy of computed tomography perfusion in patients with acute stroke: A meta-analysis

PURPOSE

The purpose of this meta-analysis was to evaluate the sensitivity and specificity of computed tomography perfusion (CTP) in diagnosing acute ischemic stroke in patients presenting to the emergency department with stroke-like symptoms.

METHODS

Medline, Cochrane, EMBASE, and Google Scholar databases were searched until November 5, 2014 using the following terms: magnetic resonance imaging/MRI, computed tomography/CT, and stroke. Randomized controlled trials, retrospective, and case-controlled studies were included which evaluated patients who presented for emergency assessment of stroke-like systems. Diffusion weighted imaging (DWI) was used as reference standard. Only studies published in English or Chinese were included. Quality assessment and sensitivity analysis were performed to evaluate that strength of the data.

RESULTS

The analysis included six studies with a total of 1429 patients. The pooled overall sensitivity for CTP indicated it had reasonable sensitivity (55.7%) and high specificity (92%). Subgroup analysis indicated that of the different CTP modes, MTT and CBF had higher sensitivities (48.6% and 47.3%, respectively) than CBV (26.3%). CBF and CBV had higher specificity (91.0% and 95.4%, respectively) compared with MTT (86.6%).

CONCLUSION

All three CTP modes had adequate sensitivity but very high specificity, and among the three CTP modes, CBF had the best diagnostic characteristics.

A CAD System for Hemorrhagic Stroke

Computer-aided detection/diagnosis (CAD) is a key component of routine clinical practice, increasingly used for detection, interpretation, quantification and decision support. Despite a critical need, there is no clinically accepted CAD system for stroke yet. Here we introduce a CAD system for hemorrhagic stroke. This CAD system segments, quantifies, and displays hematoma in 2D/3D, and supports evacuation of hemorrhage by thrombolytic treatment monitoring progression and quantifying clot removal. It supports seven-step workflow: select patient, add a new study, process patient's scans, show segmentation results, plot hematoma volumes, show 3D synchronized time series hematomas, and generate report. The system architecture contains four components: library, tools, application with user interface, and hematoma segmentation algorithm. The tools include a contour editor, 3D surface modeler, 3D volume measure, histogramming, hematoma volume plot, and 3D synchronized time-series hematoma display. The CAD system has been designed and implemented in C++. It has also been employed in the CLEAR and MISTIE phase-III, multicenter clinical trials. This stroke CAD system is potentially useful in research and clinical applications, particularly for clinical trials.

Population-based Stroke Atlas for outcome prediction: method and preliminary results for ischemic stroke from CT

BACKGROUND AND PURPOSE

Knowledge of outcome prediction is important in stroke management. We propose a lesion size and location-driven method for stroke outcome prediction using a Population-based Stroke Atlas (PSA) linking neurological parameters with neuroimaging in population. The PSA aggregates data from previously treated patients and applies them to currently treated patients. The PSA parameter distribution in the infarct region of a treated patient enables prediction. We introduce a method for PSA calculation, quantify its performance, and use it to illustrate ischemic stroke outcome prediction of modified Rankin Scale (mRS) and Barthel Index (BI).

METHODS

The preliminary PSA was constructed from 128 ischemic stroke cases calculated for 8 variants (various data aggregation schemes) and 3 case selection variables (infarct volume, NIHSS at admission, and NIHSS at day 7), each in 4 ranges. Outcome prediction for 9 parameters (mRS at 7th, and mRS and BI at 30th, 90th, 180th, 360th day) was studied using a leave-one-out approach, requiring 589,824 PSA maps to be analyzed.

RESULTS

Outcomes predicted for different PSA variants are statistically equivalent, so the simplest and most efficient variant aiming at parameter averaging is employed. This variant allows the PSA to be pre-calculated before prediction. The PSA constrained by infarct volume and NIHSS reduces the average prediction error (absolute difference between the predicted and actual values) by a fraction of 0.796; the use of 3 patient-specific variables further lowers it by 0.538. The PSA-based prediction error for mild and severe outcomes (mRS  =  [2]-[5]) is (0.5-0.7). Prediction takes about 8 seconds.

CONCLUSIONS

PSA-based prediction of individual and group mRS and BI scores over time is feasible, fast and simple, but its clinical usefulness requires further studies. The case selection operation improves PSA predictability. A multiplicity of PSAs can be computed independently for different datasets at various centers and easily merged, which enables building powerful PSAs over the community.

Characterization of intraventricular and intracerebral hematomas in non-contrast CT

Characterization of hematomas is essential in scan reading, manual delineation, and designing automatic segmentation algorithms. Our purpose is to characterize the distribution of intraventricular (IVH) and intracerebral hematomas (ICH) in NCCT scans, study their relationship to gray matter (GM), and to introduce a new tool for quantitative hematoma delineation. We used 289 serial retrospective scans of 51 patients. Hematomas were manually delineated in a two-stage process. Hematoma contours generated in the first stage were quantified and enhanced in the second stage. Delineation was based on new quantitative rules and hematoma profiling, and assisted by a dedicated tool superimposing quantitative information on scans with 3D hematoma display. The tool provides: density maps (40-85HU), contrast maps (8/15HU), mean horizontal/vertical contrasts for hematoma contours, and hematoma contours below a specified mean contrast (8HU). White matter (WM) and GM were segmented automatically. IVH/ICH on serial NCCT is characterized by 59.0HU mean, 60.0HU median, 11.6HU standard deviation, 23.9HU mean contrast, -0.99HU/day slope, and -0.24 skewness (changing over time from negative to positive). Its 0.1(st)-99.9(th) percentile range corresponds to 25-88HU range. WM and GM are highly correlated (R (2)=0.88; p<10(-10)) whereas the GM-GS correlation is weak (R (2)=0.14; p<10(-10)). The intersection point of mean GM-hematoma density distributions is at 55.6±5.8HU with the corresponding GM/hematoma percentiles of 88(th)/40(th). Objective characterization of IVH/ICH and stating the rules quantitatively will aid raters to delineate hematomas more robustly and facilitate designing algorithms for automatic hematoma segmentation. Our two-stage process is general and potentially applicable to delineate other pathologies on various modalities more robustly and quantitatively.