Prediction of hemorrhagic transformation after acute thrombolysis following major artery occlusion using relative ADC ratio: A retrospective study

Artificial intelligence/machine learning for neuroimaging to predict hemorrhagic transformation: Systematic review/meta-analysis

BACKGROUND AND PURPOSE

Early and reliable prediction of hemorrhagic transformation (HT) in patients with acute ischemic stroke (AIS) is crucial for treatment decisions and early intervention. The purpose of this study was to conduct a systematic review and meta-analysis on the performance of artificial intelligence (AI) and machine learning (ML) models that utilize neuroimaging to predict HT.

METHODS

A systematic search of PubMed, EMBASE, and Web of Science was conducted until February 19, 2024. Inclusion criteria were as follows: patients with AIS who received reperfusion therapy; AI/ML algorithm using imaging to predict HT; or presence of sufficient data on the predictive performance. Exclusion criteria were as follows: articles with less than 20 patients; articles lacking algorithms that operate solely on images; or articles not detailing the algorithm used. The quality of eligible studies was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 and Checklist for Artificial Intelligence in Medical Imaging. Pooled sensitivity, specificity, and diagnostic odds ratio (DOR) were calculated using a random-effects model, and a summary receiver operating characteristic curve was constructed using the Reitsma method.

RESULTS

We identified six eligible studies, which included 1640 patients. Aside from an unclear risk of bias regarding flow and timing identified in two of the studies, all studies showed low risk of bias and applicability concerns in all categories. Pooled sensitivity, specificity, and DOR were .849, .878, and 45.598, respectively.

CONCLUSION

AI/ML models can reliably predict the occurrence of HT in AIS patients. More prospective studies are needed for subgroup analyses and higher clinical certainty and usefulness.

Magnetic Resonance Perfusion-Weighted Imaging in Predicting Hemorrhagic Transformation of Acute Ischemic Stroke: A Retrospective Study

Hemorrhagic transformation (HT) is one of the common complications in patients with acute ischemic stroke (AIS). This study aims to investigate the value of different thresholds of Tmax generated from perfusion-weighted MR imaging (PWI) and the apparent diffusion coefficient (ADC) value in the prediction of HT in AIS. A total of 156 AIS patients were enrolled in this study, with 55 patients in the HT group and 101 patients in non-HT group. The clinical baseline data and multi-parametric MRI findings were compared between HT and non-HT groups to identify indicators related to HT. The optimal parameters for predicting HT and the corresponding cutoff values were obtained using the receiver operating characteristic curve analysis of the volumes of ADC < 620 × 10-6 mm2/s and Tmax > 6 s, 8 s, and 10 s. The results showed that the volumes of ADC < 620 × 10-6 mm2/s and Tmax > 6 s, 8 s, and 10 s in the HT group were all significantly larger than that in the non-HT group and were all independent risk factors for HT. Early measurement of the volume of Tmax > 10 s had the highest value, with a cutoff lesion volume of 10.5 mL.

A clinical-radiomics model based on noncontrast computed tomography to predict hemorrhagic transformation after stroke by machine learning: a multicenter study

OBJECTIVE

To build a clinical-radiomics model based on noncontrast computed tomography images to identify the risk of hemorrhagic transformation (HT) in patients with acute ischemic stroke (AIS) following intravenous thrombolysis (IVT).

MATERIALS AND METHODS

A total of 517 consecutive patients with AIS were screened for inclusion. Datasets from six hospitals were randomly divided into a training cohort and an internal cohort with an 8:2 ratio. The dataset of the seventh hospital was used for an independent external verification. The best dimensionality reduction method to choose features and the best machine learning (ML) algorithm to develop a model were selected. Then, the clinical, radiomics and clinical-radiomics models were developed. Finally, the performance of the models was measured using the area under the receiver operating characteristic curve (AUC).

RESULTS

Of 517 from seven hospitals, 249 (48%) had HT. The best method for choosing features was recursive feature elimination, and the best ML algorithm to build models was extreme gradient boosting. In distinguishing patients with HT, the AUC of the clinical model was 0.898 (95% CI 0.873-0.921) in the internal validation cohort, and 0.911 (95% CI 0.891-0.928) in the external validation cohort; the AUC of radiomics model was 0.922 (95% CI 0.896-0.941) and 0.883 (95% CI 0.851-0.902), while the AUC of clinical-radiomics model was 0.950 (95% CI 0.925-0.967) and 0.942 (95% CI 0.927-0.958) respectively.

CONCLUSION

The proposed clinical-radiomics model is a dependable approach that could provide risk assessment of HT for patients who receive IVT after stroke.

Predictive Value of CT Perfusion in Hemorrhagic Transformation after Acute Ischemic Stroke: A Systematic Review and Meta-Analysis

Background: Existing studies indicate that some computed tomography perfusion (CTP) parameters may predict hemorrhagic transformation (HT) after acute ischemic stroke (AIS), but there is an inconsistency in the conclusions alongside a lack of comprehensive comparison. Objective: To comprehensively evaluate the predictive value of CTP parameters in HT after AIS. Data sources: A systematical literature review of existing studies was conducted up to 1st October 2022 in six mainstream databases that included original data on the CTP parameters of HT and non-HT groups or on the diagnostic performance of relative cerebral blood flow (rCBF), relative permeability-surface area product (rPS), or relative cerebral blood volume (rCBV) in patients with AIS that completed CTP within 24 h of onset. Data Synthesis: Eighteen observational studies were included. HT and non-HT groups had statistically significant differences in CBF, CBV, PS, rCBF, rCBV, and rPS (p < 0.05 for all). The hierarchical summary receiver operating characteristic (HSROC) revealed that rCBF (area under the curve (AUC) = 0.9), rPS (AUC = 0.89), and rCBV (AUC = 0.85) had moderate diagnostic performances in predicting HT. The pooled sensitivity and specificity of rCBF were 0.85 (95% CI, 0.75−0.91) and 0.83 (95% CI, 0.63−0.94), respectively. Conclusions: rCBF, rPS, and rCBV had moderate diagnostic performances in predicting HT, and rCBF had the best pooled sensitivity and specificity.

Predictive value of magnetic resonance imaging-based texture analysis for hemorrhage transformation in large cerebral infarction

Massive cerebral infarction (MCI) is a devastating condition and associated with high rate of morbidity and mortality. Hemorrhagic transformation (HT) is a common complication after acute MCI, and often results in poor outcomes. Although several predictors of HT have been identified in acute ischemic stroke (AIS), the association between the predictors and HT remains controversial. Therefore, we aim to explore the value of texture analysis on magnetic resonance image (MRI) for predicting HT after acute MCI. This retrospective study included a total of 98 consecutive patients who were admitted for acute MCI between January 2019 and October 2020. Patients were divided into the HT group (n = 44) and non-HT group (n = 54) according to the follow-up computed tomography (CT) images. A total of 11 quantitative texture features derived from images of diffusion-weighted image (DWI) or T2-weighted-Fluid-Attenuated Inversion Recovery (T2/FLAIR) were extracted for each patient. Receiver operating characteristic (ROC) analysis were performed to determine the predictive performance of textural features, with HT as the outcome measurement. There was no significant difference in the baseline demographic and clinical characteristics between the two groups. The distribution of atrial fibrillation and National Institutes of Health Stroke Scale (NIHSS) were significantly higher in patients with HT than those without HT. Among the textural parameters extracted from DWI images, six parameters, f2 (contrast), f3 (correlation), f4 (sum of squares), f5 (inverse difference moment), f10 (difference variance), and f11 (difference entropy), differs significantly between the two groups (p < 0.05). Moreover, five of six parameters (f2, f3, f5, f10, and f11) have good predictive performances of HT with the area under the ROC curve (AUC) values of 0.795, 0.779, 0.791, 0.780, and 0.797, respectively. However, the texture features f2, f3, and f10 in T2/FLAIR images were the only three significant predictors of HT in patients with acute MCI, but with a relatively low AUC values of 0.652, 0.652, and 0.670, respectively. In summary, our preliminary results showed DWI-based texture analysis has a good predictive validity for HT in patients with acute MCI. Multiparametric MRI texture analysis model should be developed to improve the prediction performance of HT following acute MCI.

A deep learning-based model for prediction of hemorrhagic transformation after stroke

Hemorrhagic transformation (HT) is one of the most serious complications after endovascular thrombectomy (EVT) in acute ischemic stroke (AIS) patients. The purpose of this study is to develop and validate deep-learning (DL) models based on multiparametric magnetic resonance imaging (MRI) to automatically predict HT in AIS patients. Multiparametric MRI and clinical data of AIS patients with EVT from two centers (data set 1 for training and testing: n = 338; data set 2 for validating: n = 54) were used in the DL models. The acute infarction area of diffusion-weighted imaging (DWI) and hypoperfusion of perfusion-weighted imaging (PWI) was labeled manually. Two forms of data sets (volume of interest [VOI] data sets and slice data sets) were analyzed, respectively. The models based on single parameter and multiparameter models were developed and validated to predict HT in AIS patients after EVT. Performance was evaluated by area under the receiver-operating characteristic curve (AUC), accuracy (ACC), sensitivity, specificity, negative predictive value, and positive predictive value. The results showed that the performance of single parameter model based on MTT (VOI data set: AUC = 0.933, ACC = 0.843; slice data set: AUC = 0.945, ACC = 0.833) and TTP (VOI data set: AUC = 0.916, ACC = 0.873; slice data set: AUC = 0.889, ACC = 0.818) were better than the other single parameter model. The multiparameter model based on DWI & MTT & TTP & Clinical (DMTC) had the best performance for predicting HT (VOI data set: AUC = 0.948, ACC = 0.892; slice data set: AUC = 0.932, ACC = 0.873). The DMTC model in the external validation set achieved similar performance with the testing set (VOI data set: AUC = 0.939, ACC = 0.884; slice data set: AUC = 0.927, ACC = 0.871) (p > 0.05). The proposed clinical, DWI, and PWI multiparameter DL model has great potential for assisting the periprocedural management in the early prediction HT of the AIS patients with EVT.

Differences between predictive factors for early neurological deterioration due to hemorrhagic and ischemic insults following intravenous recombinant tissue plasminogen activator

Early neurological deterioration (END) following intravenous recombinant tissue plasminogen activator (rt-PA) treatment is a serious clinical event that can be caused by hemorrhagic or ischemic insult. We investigated the differences in predictive factors for END due to hemorrhagic and END due to ischemic insults. Consecutive patients from four hospitals who received 0.6 mg/kg intravenous rt-PA for acute ischemic stroke were retrospectively recruited. END was defined as a National Institutes of Health Stroke Scale (NIHSS) score ≥ 4 points within 24 h compared with baseline. END was classified into those due to hemorrhagic (END_h) or ischemic (END_i) insult based on computed tomography (CT) or magnetic resonance imaging. Risk factors associated with END_h and END_i were investigated by comparison with non-END cases. A total of 744 patients (452 men, median 75 years old) were included. END was observed in 79 patients (10.6%), including 22 END_h (3.0%) and 57 END_i (7.7%), which occurred within a median of 7 h after treatment. Multivariate analyses showed that higher pretreatment NIHSS score (odds ratio [OR] 1.06, 95% confidence interval [CI] 1.00–1.13) and pretreatment with antiplatelets (OR 2.84, 95% CI 1.08–7.72) were associated with END_h. Extensive early ischemic change (Alberta Stroke Program Early CT Score ≤ 7 on CT or ≤ 6 on diffusion-weighted imaging; OR 2.80, 95% CI 1.36–5.64) and large artery occlusions (OR 3.09, 95% CI 1.53–6.57) were associated with END_i. Distinct factors were predictive for the END subtypes. These findings could help develop preventative measures for END in patients with the identified risk factors.

MRI for prediction of hemorrhagic transformation in acute ischemic stroke: a systematic review and meta-analysis

BACKGROUND

Hemorrhagic transformation increases mortality and morbidity in patients with acute ischemic stroke.

PURPOSE

The purpose of this study is to evaluate the diagnostic performance of magnetic resonance imaging (MRI) for prediction of hemorrhagic transformation in acute ischemic stroke.

MATERIAL AND METHODS

A systematic literature search of MEDLINE and EMBASE was performed up to 27 July 2018, including the search terms "acute ischemic stroke," "hemorrhagic transformation," and "MRI." Studies evaluating the diagnostic performance of MRI for prediction of hemorrhagic transformation in acute ischemic stroke were included. Diagnostic meta-analysis was conducted with a bivariate random-effects model to calculate the pooled sensitivity and specificity. Subgroup analysis was performed including studies using advanced MRI techniques including perfusion-weighted imaging, diffusion-weighted imaging, and susceptibility-weighted imaging.

RESULTS

Nine original articles with 665 patients were included. Hemorrhagic transformation is associated with high permeability, hypoperfusion, low apparent diffusion coefficient (ADC), and FLAIR hyperintensity. The pooled sensitivity was 82% (95% confidence interval [CI] 61-93) and the pooled specificity was 79% (95% CI 71-85). The area under the hierarchical summary receiver operating characteristic curve was 0.85 (95% CI 0.82-0.88). Although study heterogeneity was present in both sensitivity (I²=67.96%) and specificity (I²=78.93%), a threshold effect was confirmed. Studies using advanced MRI showed sensitivity of 92% (95% CI 70-98) and specificity of 78% (95% CI 65-87) to conventional MRI.

CONCLUSION

MRI may show moderate diagnostic performance for predicting hemorrhage in acute ischemic stroke although the clinical significance of this hemorrhage is somewhat uncertain.

The Risk of Symptomatic Intracranial Hemorrhage after Thrombolysis for Acute Stroke: Current Concepts and Perspectives

Background

Thrombolysis is the standard of treatment for acute ischemic stroke, with a time window of up to 4½ h from stroke onset. Despite the long experience with the use of recombinant tissue plasminogen activator and the adherence to protocols symptomatic intracranial hemorrhage (SICH) may occur in around 6% of cases, with high-mortality rate and poor-functional outcomes. Many patients are excluded from thrombolysis on the basis of an evaluation of known risk factors, but there are other less known factors involved.

Objective

The purpose of this work is to analyze the less known risk factors for SICH after thrombolysis. A search of articles related with this field has been undertaken in PubMed with the keywords (brain hemorrhage, thrombolysis, and acute ischemic stroke). Some risk factors for SICH have emerged such as previous microbleeds on brain magnetic resonance imaging, leukoaraiosis, and previous antiplatelet drug use or statin use. Serum matrix metalloproteinases have emerged as a promising biomarker for better selection of patients, but further research is needed.

Conclusions

In addition to the already known risk factors considered in the standard protocols, an individualized evaluation of risks is needed to minimize the risk of brain hemorrhage after thrombolysis for ischemic stroke.