Radiomics-Based Intracranial Thrombus Features on CT and CTA Predict Recanalization with Intravenous Alteplase in Patients with Acute Ischemic Stroke

Radiomics of intrathrombus and perithrombus regions for Post-EVT intracranial hemorrhage risk Prediction: A multicenter CT study

OBJECTIVES

This study aimed to assess the predictive performance of radiomics derived from computed tomography (CT) images of thrombus regions in predicting the risk of intracranial hemorrhage (ICH) following endovascular thrombectomy (EVT).

MATERIALS AND METHODS

This retrospective multicenter study included 336 patients who underwent admission CT and EVT for acute anterior-circulation large vessel occlusion between December 2018 and December 2023. Follow-up imaging was performed 24 h post-procedure to evaluate the occurrence of ICH. 230 patients from centers A and B were randomly allocated into training and test groups in a 7:3 ratio, while the remaining 106 patients from center C comprised the validation cohort. Radiologists manually segmenting the thrombus on CT images, and the perithrombus region was defined by expanding the initial region of interest (ROI). A total of 428 radiomics features were extracted from both intrathrombus and perithrombus regions on CT images. The Mann-Whitney U test was used for feature selection, and least absolute shrinkage and selection operator (LASSO) regression was employed for model development, followed by validation using a 5-fold cross-validation approach. Model performance was assessed using the area under the curve (AUC) of the receiver operating characteristic (ROC).

RESULTS

Among the eligible patients, 128 (38.1 %) experienced ICH after EVT. The combined model exhibited superior performance in the training cohort (AUC: 0.913, 95 % CI: 0.861-0.965), test cohort (AUC: 0.868, 95 % CI: 0.775-0.962), and validation cohort (AUC: 0.850, 95 % CI: 0.768-0.912). Notably, in the validation group, both the perithrombus and combined models demonstrated higher predictive accuracy compared to the intrathrombus model (0.837 vs. 0.684, p = 0.02; AUC: 0.850 vs. 0.684, p = 0.01).

CONCLUSIONS

Radiomics features derived from the perithrombus region significantly enhance the prediction of ICH after EVT, providing valuable insights for optimizing post-procedural clinical decisions.

CLINICAL RELEVANCE STATEMENT

This study highlights the importance of radiomics extracted from intrathrombus and perithrombus region in predicting intracranial hemorrhagefollowing endovascular thrombectomy, which can aid in improving patient outcomes.

A Deep Learning Approach to Predict Recanalization First-Pass Effect following Mechanical Thrombectomy in Patients with Acute Ischemic Stroke

BACKGROUND AND PURPOSE

Following endovascular thrombectomy in patients with large-vessel occlusion stroke, successful recanalization from 1 attempt, known as the first-pass effect, has correlated favorably with long-term outcomes. Pretreatment imaging may contain information that can be used to predict the first-pass effect. Recently, applications of machine learning models have shown promising results in predicting recanalization outcomes, albeit requiring manual segmentation. In this study, we sought to construct completely automated methods using deep learning to predict the first-pass effect from pretreatment CT and MR imaging.

MATERIALS AND METHODS

Our models were developed and evaluated using a cohort of 326 patients who underwent endovascular thrombectomy at UCLA Ronald Reagan Medical Center from 2014 to 2021. We designed a hybrid transformer model with nonlocal and cross-attention modules to predict the first-pass effect on MR imaging and CT series.

RESULTS

The proposed method achieved a mean 0.8506 (SD, 0.0712) for cross-validation receiver operating characteristic area under the curve (ROC-AUC) on MR imaging and 0.8719 (SD, 0.0831) for cross-validation ROC-AUC on CT. When evaluated on the prospective test sets, our proposed model achieved a mean ROC-AUC of 0.7967 (SD, 0.0335) with a mean sensitivity of 0.7286 (SD, 0.1849) and specificity of 0.8462 (SD, 0.1216) for MR imaging and a mean ROC-AUC of 0.8051 (SD, 0.0377) with a mean sensitivity of 0.8615 (SD, 0.1131) and specificity 0.7500 (SD, 0.1054) for CT, respectively, representing the first classification of the first-pass effect from MR imaging alone and the first automated first-pass effect classification method in CT.

CONCLUSIONS

Results illustrate that both nonperfusion MR imaging and CT from admission contain signals that can predict a successful first-pass effect following endovascular thrombectomy using our deep learning methods without requiring time-intensive manual segmentation.

Deep learning prediction of stroke thrombus red blood cell content from multiparametric MRI

BACKGROUND AND PURPOSE

Thrombus red blood cell (RBC) content has been shown to be a significant factor influencing the efficacy of acute ischemic stroke treatment. In this study, our objective was to evaluate the ability of convolutional neural networks (CNNs) to predict ischemic stroke thrombus RBC content using multiparametric MR images.

MATERIALS AND METHODS

Retrieved stroke thrombi were scanned ex vivo using a three-dimensional multi-echo gradient echo sequence and histologically analyzed. 188 thrombus R2*, quantitative susceptibility mapping and late-echo GRE magnitude image slices were used to train and test a 3-layer CNN through cross-validation. Data augmentation techniques involving input equalization and random image transformation were employed to improve network performance. The network was assessed for its ability to quantitatively predict RBC content and to classify thrombi into RBC-rich and RBC-poor groups.

RESULTS

The CNN predicted thrombus RBC content with an accuracy of 62% (95% CI 48-76%) when trained on the original dataset and improved to 72% (95% CI 60-84%) on the augmented dataset. The network classified thrombi as RBC-rich or poor with an accuracy of 71% (95% CI 58-84%) and an area under the curve of 0.72 (95% CI 0.57-0.87) when trained on the original dataset and improved to 80% (95% CI 69-91%) and 0.84 (95% CI 0.73-0.95), respectively, on the augmented dataset.

CONCLUSIONS

The CNN was able to accurately predict thrombus RBC content using multiparametric MR images, and could provide a means to guide treatment strategy in acute ischemic stroke.

Radiomics Studies on Ischemic Stroke and Carotid Atherosclerotic Disease: A Reporting Quality Assessment

Objective: To assess the reporting quality of radiomics studies on ischemic stroke, intracranial and carotid atherosclerotic disease using the Image Biomarker Standardization Initiative (IBSI) reporting guidelines with the aim of finding avenues of improvement for future publications. Method: PubMed database was searched to identify relevant radiomics studies. Of 560 articles, 41 original research articles were included in this analysis. Based on IBSI radiomics reporting guidelines, checklists for CT-based and MRI-based studies were created to allow a structured and comprehensive evaluation of each study's adherence to these guidelines. Results: The main topics covered included radiomics studies were ischemic stroke, intracranial artery disease, and carotid atherosclerotic disease. The reporting checklist median score was 17/40 for the 20 CT-based radiomics studies and 22.5/50 for the 20 MRI-based studies. Basic items like imaging modality, region of interest, and image biomarker set utilized were included in all studies. However, details regarding image acquisition and reconstruction, post-acquisition image processing, and image biomarkers computation were inconsistently detailed across studies. Conclusion: The overall reporting quality of the included radiomics studies was suboptimal. These findings underscore a pressing need for improved reporting practices in radiomics research, to ensure validation and reproducibility of results. Our study provides insights into current reporting standards and highlights specific areas where adherence to IBSI guidelines could be significantly improved.

Non-contrast CT radiomics-clinical machine learning model for futile recanalization after endovascular treatment in anterior circulation acute ischemic stroke

OBJECTIVE

To establish a machine learning model based on radiomics and clinical features derived from non-contrast CT to predict futile recanalization (FR) in patients with anterior circulation acute ischemic stroke (AIS) undergoing endovascular treatment.

METHODS

A retrospective analysis was conducted on 174 patients who underwent endovascular treatment for acute anterior circulation ischemic stroke between January 2020 and December 2023. FR was defined as successful recanalization but poor prognosis at 90 days (modified Rankin Scale, mRS 4-6). Radiomic features were extracted from non-contrast CT and selected using the least absolute shrinkage and selection operator (LASSO) regression method. Logistic regression (LR) model was used to build models based on radiomic and clinical features. A radiomics-clinical nomogram model was developed, and the predictive performance of the models was evaluated using area under the curve (AUC), accuracy, sensitivity, and specificity.

RESULTS

A total of 174 patients were included. 2016 radiomic features were extracted from non-contrast CT, and 9 features were selected to build the radiomics model. Univariate and stepwise multivariate analyses identified admission NIHSS score, hemorrhagic transformation, NLR, and admission blood glucose as independent factors for building the clinical model. The AUC of the radiomics-clinical nomogram model in the training and testing cohorts were 0.860 (95%CI 0.801-0.919) and 0.775 (95%CI 0.605-0.945), respectively.

CONCLUSION

The radiomics-clinical nomogram model based on non-contrast CT demonstrated satisfactory performance in predicting futile recanalization in patients with anterior circulation acute ischemic stroke.

Hybrid Clot Histomic-Transcriptomic Models Predict Functional Outcome After Mechanical Thrombectomy in Acute Ischemic Stroke

BACKGROUND AND OBJECTIVES

Histologic and transcriptomic analyses of retrieved stroke clots have identified features associated with patient outcomes. Previous studies have demonstrated the predictive capacity of histology or expression features in isolation. Few studies, however, have investigated how paired histologic image features and expression patterns from the retrieved clots can improve understanding of clot pathobiology and our ability to predict long-term prognosis. We hypothesized that computational models trained using clot histomics and mRNA expression can predict early neurological improvement (ENI) and 90-day functional outcome (modified Rankin Scale Score, mRS) better than models developed using histological composition or expression data alone.

METHODS

We performed paired histological and transcriptomic analysis of 32 stroke clots. ENI was defined as a delta-National Institutes of Health Stroke Score/Scale > 4, and a good long-term outcome was defined as mRS ≤2 at 90 days after procedure. Clots were H&E-stained and whole-slide imaged at 40×. An established digital pathology pipeline was used to extract 237 histomic features and to compute clot percent composition (%Comp). When dichotomized by either the ENI or mRS thresholds, differentially expressed genes were identified as those with absolute fold-change >1.5 and q < 0.05. Machine learning with recursive feature elimination (RFE) was used to select clot features and evaluate computational models for outcome prognostication.

RESULTS

For ENI, RFE identified 9 optimal histologic and transcriptomic features for the hybrid model, which achieved an accuracy of 90.8% (area under the curve [AUC] = 0.98 ± 0.08) in testing and outperformed models based on histomics (AUC = 0.94 ± 0.09), transcriptomics (AUC = 0.86 ± 0.16), or %Comp (AUC = 0.70 ± 0.15) alone. For mRS, RFE identified 7 optimal histomic and transcriptomic features for the hybrid model. This model achieved an accuracy of 93.7% (AUC = 0.94 ± 0.09) in testing, also outperforming models based on histomics (AUC = 0.90 ± 0.11), transcriptomics (AUC = 0.55 ± 0.27), or %Comp (AUC = 0.58 ± 0.16) alone.

CONCLUSION

Hybrid models offer improved outcome prognostication for patients with stroke. Identified digital histology and mRNA signatures warrant further investigation as biomarkers of patient functional outcome after thrombectomy.

Application of radiomics in ischemic stroke

In recent years, radiomics has emerged as a novel research methodology that plays a crucial role in the diagnosis and treatment of ischemic stroke. By integrating multimodal medical imaging techniques such as computed tomography and magnetic resonance imaging, radiomics offers in-depth insights into aspects such as the extent of brain tissue damage and hemodynamics. These data help physicians to accurately assess patient condition, select optimal treatment strategies, and predict recovery trajectories and long-term prognoses, thereby enhancing treatment efficacy and reducing the risk of complications. With the anticipated further advancements in radiomic technology, this methodology has great potential for expanded applications in the early detection, treatment, and prognosis of ischemic stroke. The present narrative review explores the burgeoning field of radiomics and its transformative impact on ischemic stroke.

Computed tomography-based radiomics model to predict adverse clinical outcomes in acute pulmonary embolism

This preliminary study investigated the feasibility of a combined model constructed using radiomic features based on computed tomography (CT) and clinical features to predict adverse clinical outcomes in acute pulmonary embolism (APE). Currently, there is no widely recognized predictive model. Patients with confirmed APE who underwent CT pulmonary angiography were retrospectively categorized into good and poor prognosis groups. Seventy-four patients were randomized into a training (n = 51) or validation (n = 23) cohort. Feature extraction was performed using 3D-Slicer software. The least absolute shrinkage and selection operator regression was used to identify the optimal radiomics features and calculate the radiomics scores; subsequently, the radiomics model was developed. A combined predictive model was constructed based on radiomics scores and selected clinical features. The predictive efficacy of the three models (radiomics, clinical and combined) was assessed by plotting receiver operating characteristic curves. Furthermore, the calibration curves were graphed and the decision curve analysis was performed. Four radiomic features were screened to calculate the radiomic score. Right ventricular to left ventricular ratio (RV/LV) ≥ 1.0 and radiomics score were independent risk factors for adverse clinical outcomes. In the training and validation cohorts, the areas under the curve (AUCs) for the RV/LV ≥ 1.0 (clinical) and radiomics score prediction models were 0.778 and 0.833 and 0.907 and 0.817, respectively. The AUCs for the combined model of RV/LV ≥ 1.0 and radiomics score were 0.925 and 0.917, respectively. The combined and radiomics models had high clinical assessment efficacy for predicting adverse clinical outcomes in APE, demonstrating the clinical utility of both models. Calibration curves exhibited a strong level of consistency between the predictive and observed probabilities of poor and good prognoses in the combined model. The combined model of RV/LV ≥ 1.0 and radiomics score based on CT could accurately and non-invasively predict adverse clinical outcomes in patients with APE.

Value of Automatically Derived Full Thrombus Characteristics: An Explorative Study of Their Associations with Outcomes in Ischemic Stroke Patients

(1) Background: For acute ischemic strokes caused by large vessel occlusion, manually assessed thrombus volume and perviousness have been associated with treatment outcomes. However, the manual assessment of these characteristics is time-consuming and subject to inter-observer bias. Alternatively, a recently introduced fully automated deep learning-based algorithm can be used to consistently estimate full thrombus characteristics. Here, we exploratively assess the value of these novel biomarkers in terms of their association with stroke outcomes. (2) Methods: We studied two applications of automated full thrombus characterization as follows: one in a randomized trial, MR CLEAN-NO IV (n = 314), and another in a Dutch nationwide registry, MR CLEAN Registry (n = 1839). We used an automatic pipeline to determine the thrombus volume, perviousness, density, and heterogeneity. We assessed their relationship with the functional outcome defined as the modified Rankin Scale (mRS) at 90 days and two technical success measures as follows: successful final reperfusion, which is defined as an eTICI score of 2b-3, and successful first-pass reperfusion (FPS). (3) Results: Higher perviousness was significantly related to a better mRS in both MR CLEAN-NO IV and the MR CLEAN Registry. A lower thrombus volume and lower heterogeneity were only significantly related to better mRS scores in the MR CLEAN Registry. Only lower thrombus heterogeneity was significantly related to technical success; it was significantly related to a higher chance of FPS in the MR CLEAN-NO IV trial (OR = 0.55, 95% CI: 0.31-0.98) and successful reperfusion in the MR CLEAN Registry (OR = 0.88, 95% CI: 0.78-0.99). (4) Conclusions: Thrombus characteristics derived from automatic entire thrombus segmentations are significantly related to stroke outcomes.

Current status and quality of radiomics studies for predicting outcome in acute ischemic stroke patients: a systematic review and meta-analysis

Background

Pre-treatment prediction of reperfusion and long-term prognosis in acute ischemic stroke (AIS) patients is crucial for effective treatment and decision-making. Recent studies have demonstrated that the inclusion of radiomics data can improve the performance of predictive models. This paper reviews published studies focused on radiomics-based prediction of reperfusion and long-term prognosis in AIS patients.

Methods

We systematically searched PubMed, Web of Science, and Cochrane databases up to September 9, 2023, for studies on radiomics-based prediction of AIS patient outcomes. The methodological quality of the included studies was evaluated using the phase classification criteria, the radiomics quality scoring (RQS) tool, and the Prediction model Risk Of Bias Assessment Tool (PROBAST). Two separate meta-analyses were performed of these studies that predict long-term prognosis and reperfusion in AIS patients.

Results

Sixteen studies with sample sizes ranging from 67 to 3,001 were identified. Ten studies were classified as phase II, and the remaining were categorized as phase 0 (n = 2), phase I (n = 1), and phase III (n = 3). The mean RQS score of all studies was 39.41%, ranging from 5.56 to 75%. Most studies (87.5%, 14/16) were at high risk of bias due to their retrospective design. The remaining two studies were categorized as low risk and unclear risk, respectively. The pooled area under the curve (AUC) was 0.88 [95% confidence interval (CI) 0.84-0.92] for predicting the long-term prognosis and 0.80 (95% CI 0.74-0.86) for predicting reperfusion in AIS.

Conclusion

Radiomics has the potential to predict immediate reperfusion and long-term outcomes in AIS patients. Further external validation and evaluation within the clinical workflow can facilitate personalized treatment for AIS patients. This systematic review provides valuable insights for optimizing radiomics prediction systems for both reperfusion and long-term outcomes in AIS patients.

Systematic review registration

https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023461671, identifier CRD42023461671.

Multimodal CT imaging of ischemic stroke thrombi identifies scale-invariant radiomic features that reflect clot biology

BACKGROUND

Biological interpretability of ischemic stroke clot imaging remains challenging.

OBJECTIVE

To carry out paired CT/micro-CT imaging of ischemic stroke clots retrieved by thrombectomy with the aim of identifying interpretable image features that are correlated among pretreatment image modalities and post-treatment histopathology.

METHODS

We performed multimodal CT imaging and histology for 10 stroke clots retrieved by mechanical thrombectomy. Clots were manually segmented from co-registered, pretreatment CT angiography (CTA) and non-contrast CT (NCCT). For the same cases, retrieved clots were iodine-stained, and imaged with a ScanCo micro-CT 100 (4.9 µm resolution). Afterwards, clots were subjected to histological processing (hematoxylin and eosin staining) and whole slide scanned (40X). Clot radiomic features (RFs) (n=93 per modality, 279 total) were extracted using PyRadiomics and histological composition was computed using Orbit Image Analysis. Correlation analysis was used to test associations between micro-CT and CTA (or NCCT) RFs as well as between RFs and histological composition. Statistical significance was considered at R≥0.65 and q<0.05.

RESULTS

From paired RF correlation analysis, we identified 23 scale-invariant RFs with significant correlation between micro-CT and CTA (18), and micro-CT and NCCT (5). Correlation of unpaired RFs identified 377 positively and 36 negatively correlated RFs between micro-CT and CTA, and 168 positively and 41 negatively correlated RFs between micro-CT and NCCT. Scale-invariant RFs computed from CTA and NCCT demonstrated significant correlation with red blood cell and fibrin-platelet components, while micro-CT RFs were found to be correlated with white blood cell percent composition.

CONCLUSION

Multimodal CT, radiomic, and histological analysis of stroke clots can help to bridge the gap between pretreatment imaging and clot pathobiology.

High-resolution magnetic resonance imaging-based radiomic features aid in selecting endovascular candidates among patients with cerebral venous sinus thrombosis

OBJECTIVES

Cerebral venous sinus thrombosis (CVST) can cause sinus obstruction and stenosis, with potentially fatal consequences. High-resolution magnetic resonance imaging (HRMRI) can diagnose CVST qualitatively, although quantitative screening methods are lacking for patients refractory to anticoagulation therapy and who may benefit from endovascular treatment (EVT). Thus, in this study, we used radiomic features (RFs) extracted from HRMRI to build machine learning models to predict response to drug therapy and determine the appropriateness of EVT.

MATERIALS AND METHODS

RFs were extracted from three-dimensional T1-weighted motion-sensitized driven equilibrium (MSDE), T2-weighted MSDE, T1-contrast, and T1-contrast MSDE sequences to build radiomic signatures and support vector machine (SVM) models for predicting the efficacy of standard drug therapy and the necessity of EVT.

RESULTS

We retrospectively included 53 patients with CVST in a prospective cohort study, among whom 14 underwent EVT after standard drug therapy failed. Thirteen RFs were selected to construct the RF signature and CVST-SVM models. In the validation dataset, the sensitivity, specificity, and area under the curve performance for the RF signature model were 0.833, 0.937, and 0.977, respectively. The radiomic score was correlated with days from symptom onset, history of dyslipidemia, smoking, fibrin degradation product, and D-dimer levels. The sensitivity, specificity, and area under the curve for the CVST-SVM model in the validation set were 0.917, 0.969, and 0.992, respectively.

CONCLUSIONS

The CVST-SVM model trained with RFs extracted from HRMRI outperformed the RF signature model and could aid physicians in predicting patient responses to drug treatment and identifying those who may require EVT.

Radiomics features on computed tomography reflect thrombus histological age prior to endovascular treatment of acute ischemic stroke

PURPOSE

To investigate the role of radiomics features in thrombus age identification and establish a CT-based radiomics model for predicting thrombus age of large vessel occlusion stroke patients.

METHODS

We retrospectively reviewed patients with middle cerebral artery occlusion receiving mechanical thrombectomy from July 2020 to March 2022 at our center. The retrieved clots were stained with Hematoxylin and Eosin (H&E) and determined as fresh or older thrombi based on coagulation age. Clot-derived radiomics features were selected by least absolute shrinkage and selection operator (LASSO) regression analysis, by which selected radiomics features were integrated into the Rad-score via the corresponding coefficients. The prediction performance of Rad-score in thrombus age was evaluated with the area under the curve (AUC) of receiver operating characteristic (ROC) curve analysis.

RESULTS

A total of 104 patients were included in our analysis, with 52 in training and 52 in validation cohort. Older thrombi were characterized with delayed procedure time, worse functional outcome and marginally associated with more attempts of device. We extracted 982 features from NCCT images. Following T test and LASSO analysis in training cohort, six radiomics features were selected, based on which the Rad-score was generated by the linear combination of features. The Rad-score showed satisfactory performance in distinguishing fresh with older thrombi, with the AUC of 0.873 (95 %CI: 0.777-0.956) and 0.773 (95 %CI: 0.636-0.910) in training and validation cohort, respectively.

CONCLUSION

This study established and validated a CT-based radiomics model that could accurately differentiate fresh with older thrombi for stroke patients receiving mechanical thrombectomy.

Recent developments in pre-hospital and in-hospital triage for endovascular stroke treatment

Triage describes the assignment of resources based on where they can be best used, are most needed, or are most likely to achieve success. Triage is of particular importance in time-critical conditions such as acute ischemic stroke. In this setting, one of the goals of triage is to minimize the delay to endovascular thrombectomy (EVT), without delaying intravenous thrombolysis or other time-critical treatments including patients who cannot benefit from EVT. EVT triage is highly context-specific, and depends on availability of financial resources, staff resources, local infrastructure, and geography. Furthermore, the EVT triage landscape is constantly changing, as EVT indications evolve and new neuroimaging methods, EVT technologies, and adjunctive medical treatments are developed and refined. This review provides an overview of recent developments in EVT triage at both the pre-hospital and in-hospital stages. We discuss pre-hospital large vessel occlusion detection tools, transport paradigms, in-hospital workflows, acute stroke neuroimaging protocols, and angiography suite workflows. The most important factor in EVT triage, however, is teamwork. Irrespective of any new technology, EVT triage will only reach optimal performance if all team members, including paramedics, nurses, technologists, emergency physicians, neurologists, radiologists, neurosurgeons, and anesthesiologists, are involved and engaged. Thus, building sustainable relationships through continuous efforts and hands-on training forms an integral part in ensuring rapid and efficient EVT triage.

Risk Stratification for Oropharyngeal Squamous Cell Carcinoma Using Texture Analysis on CT - A Step Beyond HPV Status

Background and Purpose: Human papillomavirus-associated oropharyngeal squamous cell carcinoma (OPSCC) is increasingly prevalent. Despite the overall more favorable outcome, the observed heterogeneous treatment response within this patient group highlights the need for additional means to prognosticate and guide clinical decision-making. Promising prediction models using radiomics from primary OPSCC have been derived. However, no model/s using metastatic lymphadenopathy exist to allow prognostication in those instances when the primary tumor is not seen. The aim of our study was to evaluate whether radiomics using metastatic lymphadenopathy allows for the development of a useful risk assessment model comparable to the primary tumor and whether additional knowledge of the HPV status further improves its prognostic efficacy. Materials and Methods: 80 consecutive patients diagnosed with stage III-IV OPSCC between February 2009 and October 2015, known human papillomavirus status, and pre-treatment CT images were retrospectively identified. Manual segmentation of primary tumor and metastatic lymphadenopathy was performed and the extracted texture features were used to develop multivariate assessment models to prognosticate treatment response. Results: Texture analysis of either the primary or metastatic lymphadenopathy from pre-treatment enhanced CT images can be used to develop models for the stratification of treatment outcomes in OPSCC patients. AUCs range from .78 to .85 for the various OPSCC groups tested, indicating high predictive capability of the models. Conclusions: This preliminary study can form the basis multi-centre trial that may help optimize treatment and improve quality of life in patients with OPSCC in the era of personalized medicine.

Computed Tomography Angiography-Based Thrombus Radiomics for Predicting the Time Since Stroke Onset

RATIONALE AND OBJECTIVES

The measurement of the time since stroke onset (TSS) is crucial for decision-making in the treatment of acute ischemic stroke (AIS). This study assessed the utility of computed tomography angiography (CTA) radiomics features (RFs) to estimate TSS.

MATERIALS AND METHODS

A total of 221 patients with AIS were enrolled in this retrospective study and were divided into a training group (n = 154) and a test group (n = 67). Thrombi in CTA images were manually outlined using ITK-SNAP. Images were aligned, normalized, and pre-processed to extract RFs. The TSS was calculated as the time from stroke onset to CTA completion. The patients were classified into two groups according to estimated TSS: ≤4.5 and >4.5 hours. A total of 944 RFs were extracted from CTA images. Clinical factors associated with TSS were identified using multivariate logistic regression, and a combined model (clinical data and RFs) was constructed. The predictive value of the models was assessed by the area under the receiver operating characteristic curve (AUC). The performance of the models was compared using the DeLong test, and clinical utility was evaluated by decision curve analysis.

RESULTS

The AUC of the radiomics model was 0.803 (95% confidence interval [CI]: 0.733-0.873) and 0.803 (95% CI: 0.698-0.908) in the training and test cohorts, respectively. The AUC of the combined model (containing data on age, diabetes, and atrial fibrillation) in the training and test sets was 0.813 (95% CI: 0.750-0.889) and 0.803 (95% CI: 0.699-0.907), respectively. The DeLong test showed no significant difference between the radiomics and combined models. Decision curve analysis showed that both models had clinical utility.

CONCLUSION

CTA-based thrombus radiomics can estimate TSS in patients with AIS. The addition of clinical data to the model does not improve predictive performance.

Automated Intracranial Clot Detection: A Promising Tool for Vascular Occlusion Detection in Non-Enhanced CT

(1) Background: to test the diagnostic performance of a fully convolutional neural network-based software prototype for clot detection in intracranial arteries using non-enhanced computed tomography (NECT) imaging data. (2) Methods: we retrospectively identified 85 patients with stroke imaging and one intracranial vessel occlusion. An automated clot detection prototype computed clot location, clot length, and clot volume in NECT scans. Clot detection rates were compared to the visual assessment of the hyperdense artery sign by two neuroradiologists. CT angiography (CTA) was used as the ground truth. Additionally, NIHSS, ASPECTS, type of therapy, and TOAST were registered to assess the relationship between clinical parameters, image results, and chosen therapy. (3) Results: the overall detection rate of the software was 66%, while the human readers had lower rates of 46% and 24%, respectively. Clot detection rates of the automated software were best in the proximal middle cerebral artery (MCA) and the intracranial carotid artery (ICA) with 88-92% followed by the more distal MCA and basilar artery with 67-69%. There was a high correlation between greater clot length and interventional thrombectomy and between smaller clot length and rather conservative treatment. (4) Conclusions: the automated clot detection prototype has the potential to detect intracranial arterial thromboembolism in NECT images, particularly in the ICA and MCA. Thus, it could support radiologists in emergency settings to speed up the diagnosis of acute ischemic stroke, especially in settings where CTA is not available.

Thrombus radiomics in patients with anterior circulation acute ischemic stroke undergoing endovascular treatment

BACKGROUND

Thrombus radiomics (TR) describe complex shape and textural thrombus imaging features. We aimed to study the relationship of TR extracted from non-contrast CT with procedural and functional outcome in endovascular-treated patients with acute ischemic stroke.

METHODS

Thrombi were segmented on thin-slice non-contrast CT (≤1 mm) from 699 patients included in the MR CLEAN Registry. In a pilot study, we selected 51 TR with consistent values across two raters' segmentations (ICC >0.75). Random forest models using TR in addition or as a substitute to baseline clinical variables (CV) and manual thrombus measurements (MTM) were trained with 499 patients and evaluated on 200 patients for predicting successful reperfusion (extended Thrombolysis in Cerebral Ischemia (eTICI) ≥2B), first attempt reperfusion, reperfusion within three attempts, and functional independence (modified Rankin Scale (mRS) ≤2). Three texture and shape features were selected based on feature importance and related to eTICI ≥2B, number of attempts to eTICI ≥2B, and 90-day mRS with ordinal logistic regression.

RESULTS

Random forest models using TR, CV or MTM had comparable predictive performance. Thrombus texture (inverse difference moment normalized) was independently associated with reperfusion (adjusted common OR (acOR) 0.85, 95% CI 0.72 to 0.99). Thrombus volume and texture were also independently associated with the number of attempts to successful reperfusion (acOR 1.36, 95% CI 1.03 to 1.88 and acOR 1.24, 95% CI 1.04 to 1.49).

CONCLUSIONS

TR describing thrombus volume and texture were associated with more attempts to successful reperfusion. Compared with models using CV and MTM, TR had no added value for predicting procedural and functional outcome.

OEDL: an optimized ensemble deep learning method for the prediction of acute ischemic stroke prognoses using union features

Background

Early stroke prognosis assessments are critical for decision-making regarding therapeutic intervention. We introduced the concepts of data combination, method integration, and algorithm parallelization, aiming to build an integrated deep learning model based on a combination of clinical and radiomics features and analyze its application value in prognosis prediction.

Methods

The research steps in this study include data source and feature extraction, data processing and feature fusion, model building and optimization, model training, and so on. Using data from 441 stroke patients, clinical and radiomics features were extracted, and feature selection was performed. Clinical, radiomics, and combined features were included to construct predictive models. We applied the concept of deep integration to the joint analysis of multiple deep learning methods, used a metaheuristic algorithm to improve the parameter search efficiency, and finally, developed an acute ischemic stroke (AIS) prognosis prediction method, namely, the optimized ensemble of deep learning (OEDL) method.

Results

Among the clinical features, 17 features passed the correlation check. Among the radiomics features, 19 features were selected. In the comparison of the prediction performance of each method, the OEDL method based on the concept of ensemble optimization had the best classification performance. In the comparison to the predictive performance of each feature, the inclusion of the combined features resulted in better classification performance than that of the clinical and radiomics features. In the comparison to the prediction performance of each balanced method, SMOTEENN, which is based on a hybrid sampling method, achieved the best classification performance than that of the unbalanced, oversampled, and undersampled methods. The OEDL method with combined features and mixed sampling achieved the best classification performance, with 97.89, 95.74, 94.75, 94.03, and 94.35% for Macro-AUC, ACC, Macro-R, Macro-P, and Macro-F1, respectively, and achieved advanced performance in comparison with that of methods in previous studies.

Conclusion

The OEDL approach proposed herein could effectively achieve improved stroke prognosis prediction performance, the effect of using combined data modeling was significantly better than that of single clinical or radiomics feature models, and the proposed method had a better intervention guidance value. Our approach is beneficial for optimizing the early clinical intervention process and providing the necessary clinical decision support for personalized treatment.

Artificial Intelligence for Clinical Decision Support in Acute Ischemic Stroke: A Systematic Review

BACKGROUND

Established randomized trial-based parameters for acute ischemic stroke group patients into generic treatment groups, leading to attempts using various artificial intelligence (AI) methods to directly correlate patient characteristics to outcomes and thereby provide decision support to stroke clinicians. We review AI-based clinical decision support systems in the development stage, specifically regarding methodological robustness and constraints for clinical implementation.

METHODS

Our systematic review included full-text English language publications proposing a clinical decision support system using AI techniques for direct decision support in acute ischemic stroke cases in adult patients. We (1) describe data and outcomes used in those systems, (2) estimate the systems' benefits compared with traditional stroke diagnosis and treatment, and (3) reported concordance with reporting standards for AI in healthcare.

RESULTS

One hundred twenty-one studies met our inclusion criteria. Sixty-five were included for full extraction. In our sample, utilized data sources, methods, and reporting practices were highly heterogeneous.

CONCLUSIONS

Our results suggest significant validity threats, dissonance in reporting practices, and challenges to clinical translation. We outline practical recommendations for the successful implementation of AI research in acute ischemic stroke treatment and diagnosis.

Automated Segmentation of Intracranial Thrombus on NCCT and CTA in Patients with Acute Ischemic Stroke Using a Coarse-to-Fine Deep Learning Model

BACKGROUND AND PURPOSE

Identifying the presence and extent of intracranial thrombi is crucial in selecting patients with acute ischemic stroke for treatment. This article aims to develop an automated approach to quantify thrombus on NCCT and CTA in patients with stroke.

MATERIALS AND METHODS

A total of 499 patients with large-vessel occlusion from the Safety and Efficacy of Nerinetide in Subjects Undergoing Endovascular Thrombectomy for Stroke (ESCAPE-NA1) trial were included. All patients had thin-section NCCT and CTA images. Thrombi contoured manually were used as reference standard. A deep learning approach was developed to segment thrombi automatically. Of 499 patients, 263 and 66 patients were randomly selected to train and validate the deep learning model, respectively; the remaining 170 patients were independently used for testing. The deep learning model was quantitatively compared with the reference standard using the Dice coefficient and volumetric error. The proposed deep learning model was externally tested on 83 patients with and without large-vessel occlusion from another independent trial.

RESULTS

The developed deep learning approach obtained a Dice coefficient of 70.7% (interquartile range, 58.0%-77.8%) in the internal cohort. The predicted thrombi length and volume were correlated with those of expert-contoured thrombi (r = 0.88 and 0.87, respectively; P < .001). When the derived deep learning model was applied to the external data set, the model obtained similar results in patients with large-vessel occlusion regarding the Dice coefficient (66.8%; interquartile range, 58.5%-74.6%), thrombus length (r = 0.73), and volume (r = 0.80). The model also obtained a sensitivity of 94.12% (32/34) and a specificity of 97.96% (48/49) in classifying large-vessel occlusion versus non-large-vessel occlusion.

CONCLUSIONS

The proposed deep learning method can reliably detect and measure thrombi on NCCT and CTA in patients with acute ischemic stroke.

CT-based thrombus radiomics nomogram for predicting secondary embolization during mechanical thrombectomy for large vessel occlusion

Background and aims

Secondary embolization (SE) during mechanical thrombectomy (MT) for cerebral large vessel occlusion (LVO) could reduce the anterior blood flow and worsen clinical outcomes. The current SE prediction tools have limited accuracy. In this study, we aimed to develop a nomogram to predict SE following MT for LVO based on clinical features and radiomics extracted from computed tomography (CT) images.

Materials and methods

A total of 61 patients with LVO stroke treated by MT at Beijing Hospital were included in this retrospective study, of whom 27 developed SE during the MT procedure. The patients were randomly divided (7:3) into training (n = 42) and testing (n = 19) cohorts. The thrombus radiomics features were extracted from the pre-interventional thin-slice CT images, and the conventional clinical and radiological indicators associated with SE were recorded. A support vector machine (SVM) learning model with 5-fold cross-verification was used to obtain the radiomics and clinical signatures. For both signatures, a prediction nomogram for SE was constructed. The signatures were then combined using the logistic regression analysis to construct a combined clinical radiomics nomogram.

Results

In the training cohort, the area under the receiver operating characteristic curve (AUC) of the nomograms was 0.963 for the combined model, 0.911 for the radiomics, and 0.891 for the clinical model. Following validation, the AUCs were 0.762 for the combined model, 0.714 for the radiomics model, and 0.637 for the clinical model. The combined clinical and radiomics nomogram had the best prediction accuracy in both the training and test cohort.

Conclusion

This nomogram could be used to optimize the surgical MT procedure for LVO based on the risk of developing SE.

DWI-Based Radiomics Predicts the Functional Outcome of Endovascular Treatment in Acute Basilar Artery Occlusion

BACKGROUND AND PURPOSE

Endovascular treatment is a reference treatment for acute basilar artery occlusion (ABAO). However, no established and specific methods are available for the preoperative screening of patients with ABAO suitable for endovascular treatment. This study explores the potential value of DWI-based radiomics in predicting the functional outcomes of endovascular treatment in ABAO.

MATERIALS AND METHODS

Patients with ABAO treated with endovascular treatment from the BASILAR registry (91 patients in the training cohort) and the hospitals in the Northwest of China (31 patients for the external testing cohort) were included in this study. The Mann-Whitney U test, random forests algorithm, and least absolute shrinkage and selection operator were used to reduce the feature dimension. A machine learning model was developed on the basis of the training cohort to predict the prognosis of endovascular treatment. The performance of the model was evaluated on the independent external testing cohort.

RESULTS

A subset of radiomics features (n = 6) was used to predict the functional outcomes in patients with ABAO. The areas under the receiver operating characteristic curve of the radiomics model were 0.870 and 0.781 in the training cohort and testing cohort, respectively. The accuracy of the radiomics model was 77.4%, with a sensitivity of 78.9%, specificity of 75%, positive predictive value of 83.3%, and negative predictive value of 69.2% in the testing cohort.

CONCLUSIONS

DWI-based radiomics can predict the prognosis of endovascular treatment in patients with ABAO, hence allowing a potentially better selection of patients who are most likely to benefit from this treatment.

Quality assessment of stroke radiomics studies: Promoting clinical application

PURPOSE

To evaluate the quality of radiomics studies on stroke using a radiomics quality score (RQS), Minimum Information for Medial AI reporting (MINIMAR) and Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) to promote clinical application.

METHODS

PubMed MEDLINE and Embase were searched to identify radiomics studies on stroke. Of 464 articles, 52 relevant original research articles were included. The RQS, MINIMAR and TRIPOD were scored to evaluate the quality of the studies by neuroradiologists.

RESULTS

Only four studies (7.7 %) performed external validation. The mean RQS was 3.2 of 36 (8.9 %), and the basic adherence rate was 24.9 %. The adherence rate was low for conducting phantom study (1.9 %), stating comparison to 'gold standard' (1.9 %), offering potential clinical utility (13.5 %) and performing cost-effectiveness analysis (1.9 %). None of the studies performed a test-retest, stated biologic correlation, conducted prospective studies, or opened codes and data to the public, resulting in low RQS. The total MINIMAR adherence rate was 47.4 %. The overall adherence rate for TRIPOD was 54.6 %, with low scores for reporting the title (2.0 %), key elements of the study setting (6.1 %), and explaining the sample size (2.0 %).

CONCLUSIONS

The overall radiomics reporting quality and reporting of published radiomics studies on stoke was suboptimal. More thorough validation and open data are needed to increase clinical applicability of radiomics studies.

Histologically interpretable clot radiomic features predict treatment outcomes of mechanical thrombectomy for ischemic stroke

PURPOSE

Radiomics features (RFs) extracted from CT images may provide valuable information on the biological structure of ischemic stroke blood clots and mechanical thrombectomy outcome. Here, we aimed to identify RFs predictive of thrombectomy outcomes and use clot histomics to explore the biology and structure related to these RFs.

METHODS

We extracted 293 RFs from co-registered non-contrast CT and CTA. RFs predictive of revascularization outcomes defined by first-pass effect (FPE, near to complete clot removal in one thrombectomy pass), were selected. We then trained and cross-validated a balanced logistic regression model fivefold, to assess the RFs in outcome prediction. On a subset of cases, we performed digital histopathology on the clots and computed 227 histomic features from their whole slide images as a means to interpret the biology behind significant RF.

RESULTS

We identified 6 significantly-associated RFs. RFs reflective of continuity in lower intensities, scattered higher intensities, and intensities with abrupt changes in texture were associated with successful revascularization outcome. For FPE prediction, the multi-variate model had high performance, with AUC = 0.832 ± 0.031 and accuracy = 0.760 ± 0.059 in training, and AUC = 0.787 ± 0.115 and accuracy = 0.787 ± 0.127 in cross-validation testing. Each of the 6 RFs was related to clot component organization in terms of red blood cell and fibrin/platelet distribution. Clots with more diversity of components, with varying sizes of red blood cells and fibrin/platelet regions in the section, were associated with RFs predictive of FPE.

CONCLUSION

Upon future validation in larger datasets, clot RFs on CT imaging are potential candidate markers for FPE prediction.

MRI Radiomics and Predictive Models in Assessing Ischemic Stroke Outcome-A Systematic Review

Stroke is a leading cause of disability and mortality, resulting in substantial socio-economic burden for healthcare systems. With advances in artificial intelligence, visual image information can be processed into numerous quantitative features in an objective, repeatable and high-throughput fashion, in a process known as radiomics analysis (RA). Recently, investigators have attempted to apply RA to stroke neuroimaging in the hope of promoting personalized precision medicine. This review aimed to evaluate the role of RA as an adjuvant tool in the prognosis of disability after stroke. We conducted a systematic review following the PRISMA guidelines, searching PubMed and Embase using the keywords: 'magnetic resonance imaging (MRI)', 'radiomics', and 'stroke'. The PROBAST tool was used to assess the risk of bias. Radiomics quality score (RQS) was also applied to evaluate the methodological quality of radiomics studies. Of the 150 abstracts returned by electronic literature research, 6 studies fulfilled the inclusion criteria. Five studies evaluated predictive value for different predictive models (PMs). In all studies, the combined PMs consisting of clinical and radiomics features have achieved the best predictive performance compared to PMs based only on clinical or radiomics features, the results varying from an area under the ROC curve (AUC) of 0.80 (95% CI, 0.75-0.86) to an AUC of 0.92 (95% CI, 0.87-0.97). The median RQS of the included studies was 15, reflecting a moderate methodological quality. Assessing the risk of bias using PROBAST, potential high risk of bias in participants selection was identified. Our findings suggest that combined models integrating both clinical and advanced imaging variables seem to better predict the patients' disability outcome group (favorable outcome: modified Rankin scale (mRS) ≤ 2 and unfavorable outcome: mRS > 2) at three and six months after stroke. Although radiomics studies' findings are significant in research field, these results should be validated in multiple clinical settings in order to help clinicians to provide individual patients with optimal tailor-made treatment.

Radiomics-based intracranial thrombus features on preoperative noncontrast CT predicts successful recanalization of mechanical thrombectomy in acute ischemic stroke

Background

To evaluate the predictive value of radiomics features extracted from the thrombus on preoperative computed tomography images to identify successful recanalization after stent retrieve (SR) treatment in patients with acute ischemic stroke (AIS).

Methods

Two hundred fifty-six patients newly diagnosed AIS between March 2017 and September 2020 from two institutes, including the first affiliated hospital of Soochow university (institute I) and Northern Jiangsu People's hospital (institute II), were enrolled continuously and retrospectively. Patients with unsatisfactory image quality were excluded. The remaining patients of institute I were randomly divided into the training and internal validation cohorts at a ratio of 7 to 3, and patients of institute II were collected as the external validation cohort. After extraction and selection of the optimal radiomics features from training cohort, six machine learning (ML) classifiers including naïve Bayes (NB), random forest (RF), logistic regression (LR), linear support vector machine (L.SVM), radial SVM (R.SVM), and an artificial neural network (ANN) were developed to predict successful recanalization with SR treatment and compared. A combined model based on the optimal ML classifier was constructed using the optimal radiomics model and clinical-radiological risk variables. Finally, the performance of the model was selected based on the Matthews correlation coefficient (MCC) and the area under the receiver operating (AUC) and independently evaluated on the internal validation and external validation cohorts.

Results

We automatically extracted 1,130 radiomics features from the voxel of interest (VOI) using PyRadiomics. The eight most relevant radiomics features were identified using Intraclass coefficient, single-factor logistic regression analysis, and least absolute shrinkage and selection operator algorithm in the training cohort. Among the six ML classifiers, the ANN classifier using thrombus radiomics features achieved the best prediction of early recanalization under SR with MCCs of 0.913, 0.693 and 0.505 in training, internal and external validation cohorts, respectively. Moreover, receiver operating characteristic curves showed that the combined model [AUC =0.860, 95% confidence interval (CI): 0.731-0.936; AUC =0.849, 95% CI: 0.759-0.831] was not significantly better than radiomics model based on the ANN classifier alone (AUC =0.873, 95% CI: 0.803-0.891; AUC =0.805, 95% CI: 0.864-0.971) (P>0.05, Delong test) in internal and external validation cohorts.

Conclusions

A radiomics model based on the ANN classifier has the ability to predict successful recanalization after SR in patients with AIS, thus allowing a potentially better selection of mechanical thrombectomy treatment.

Clot-based radiomics model for cardioembolic stroke prediction with CT imaging before recanalization: a multicenter study

OBJECTIVES

To develop a clot-based radiomics model using CT imaging radiomic features and machine learning to identify cardioembolic (CE) stroke before mechanical thrombectomy (MTB) in patients with acute ischemic stroke (AIS).

MATERIALS AND METHODS

This retrospective four-center study consecutively included 403 patients with AIS who sequentially underwent CT and MTB between April 2016 and July 2021. These were grouped into training, testing, and external validation cohorts. Thrombus-extracted radiomic features and basic information were gathered to construct a machine learning model to predict CE stroke. The radiological characteristics and basic information were used to build a routine radiological model. A combined radiomics and radiological features model was also developed. The performances of all models were evaluated and compared in the validation cohort. A histological analysis helped further assess the proposed model in all patients.

RESULTS

The radiomics model yielded an area under the curve (AUC) of 0.838 (95% confidence interval [CI], 0.771-0.891) for predicting CE stroke in the validation cohort, significantly higher than the radiological model (AUC, 0.713; 95% CI, 0.636-0.781; p = 0.007) but similar to the combined model (AUC, 0.855; 95% CI, 0.791-0.906; p = 0.14). The thrombus radiomic features achieved stronger correlations with red blood cells (|r_max|, 0.74 vs. 0.32) and fibrin and platelet (|r_max|, 0.68 vs. 0.18) than radiological characteristics.

CONCLUSION

The proposed CT-based radiomics model could reliably predict CE stroke in AIS, performing better than the routine radiological method.

KEY POINTS

• Admission CT imaging could offer valuable information to identify the acute ischemic stroke source by radiomics analysis. • The proposed CT imaging-based radiomics model yielded a higher area under the curve (0.838) than the routine radiological method (0.713; p = 0.007). • Several radiomic features showed significantly stronger correlations with two main thrombus constituents (red blood cells, |r_max|, 0.74; fibrin and platelet, |r_max|, 0.68) than routine radiological characteristics.

Introduction to radiomics for a clinical audience

Radiomics is a rapidly developing field of research focused on the extraction of quantitative features from medical images, thus converting these digital images into minable, high-dimensional data, which offer unique biological information that can enhance our understanding of disease processes and provide clinical decision support. To date, most radiomics research has been focused on oncological applications; however, it is increasingly being used in a raft of other diseases. This review gives an overview of radiomics for a clinical audience, including the radiomics pipeline and the common pitfalls associated with each stage. Key studies in oncology are presented with a focus on both those that use radiomics analysis alone and those that integrate its use with other multimodal data streams. Importantly, clinical applications outside oncology are also presented. Finally, we conclude by offering a vision for radiomics research in the future, including how it might impact our practice as radiologists.

A nomogram for predicting thrombus composition in stroke patients with large vessel occlusion: combination of thrombus density and perviousness with clinical features

PURPOSE

To establish a nomogram incorporating pretreatment imaging parameters and clinical characteristics for predicting the thrombus composition of acute ischemic stroke (AIS) with large vessel occlusion (LVO).

METHODS

We retrospectively enrolled patients with occlusion of the Middle Cerebral Artery (MCA) who underwent Mechanical Thrombectomy (MT). Retrieved thrombi were stained with Hematoxylin and Eosin (H&E) and Martius Scarlet Blue (MSB). Thrombi are assigned to the Fibrin-rich or RBC-rich group based on the relative fractions of Red Blood Cells (RBC), fibrin, and platelet. The independent risk factors for Fibrin-rich clots were determined via univariate and multivariate logistic regression analysis and were then integrated to establish a nomogram.

RESULTS

In total, 98 patients were included in this study. Patients with fibrin-rich clots had worse functional outcome [modified Rankin scale (mRS) 0-2, 34.7% vs 63.2%, p = 0.005], longer procedure time (76.8 min vs 50.8 min, p = 0.001), and increased maneuvers of MT (1.84 vs 1.46, p = 0.703) than those with RBC-rich clots. The independent risk factors for Fibrin-rich clots were lower perviousness measured by Non-Contrast Computer Tomography (NCCT) and CT Angiography (CTA), lower thrombus relative attenuation on NCCT, elevated Platelet-WBC ratio (PWR) of admission peripheral blood, and previous antithrombotic medication. The nomogram showed good discrimination with an area under the Receiver Operating Characteristic (ROC) curve (AUC) of 0.852 (95% CI: 0.778-0.926). The calibration curve and decision curve analysis also displayed satisfactory accuracy and clinical utility.

CONCLUSION

This study has developed and internally validated an easy-to-use nomogram which can help predict clot composition and optimize therapeutic strategies for thrombectomy.

Spatial heterogeneity of occlusive thrombus in acute ischemic stroke: A systematic review

Following the advent of mechanical thrombectomy, occlusive clots in ischemic stroke have been amply characterized using conventional histopathology. Many studies have investigated the compositional variability of thrombi and the consequences of thrombus composition on treatment response. More recent evidence has emerged about the spatial heterogeneity of the clot or the preferential distribution of its components and compact nature. Here we review this emerging body of evidence, discuss its potential clinical implications, and propose the development of adequate characterization techniques.

Prognostic value of radiomics-based hyperdense middle cerebral artery sign for patients with acute ischemic stroke after thrombectomy strategy

Background and purpose

The purpose of this study was to evaluate the prognostic value of radiomics-based hyperdense middle cerebral artery sign (HMCAS) for patients with acute ischemic stroke (AIS) after mechanical thrombectomy (MT) and to establish prediction models to identify patients who may benefit more from MT.

Methods

In this retrospective study, a total of 102 consecutive patients who presented with HMCAS on non-contrast computed tomography (NCCT) at admission and underwent MT in our hospital between January 2019 and December 2020 were recruited. Among them, 46 experienced favorable outcomes (modified Rankin Scale [mRS] ≤ 2) at 3 months of follow-up. All patients were categorized into two sets, namely, the training set (n = 81) and the test set (n = 21). Radiomics features (RFs) and clinical features (CFs) in the training set were selected by statistical analysis to create models. The models' discriminative ability was quantified using the area under the curve (AUC) and confirmed by decision curve analyses.

Results

The prediction model established using CFs before MT includes baseline National Institutes of Health Stroke Scale (NIHSS) and neutrophil-to-lymphocyte ratio (NLR) [AUC [95% confidence interval (CI)] = 0.596 (0.312-0.881)]. A total of 1,389 RFs were extracted from each hyperdense territory and 8 RFs were left to build the radiomics model [RM; AUC (95%CI) = 0.798 (0.598-0.998)]. The model using preoperative CFs and RFs showed good performance [AUC (95%CI) = 0.817 (0.625-1.000)]. The models using post-operative CFs alone [AUC (95%CI) = 0.856 (0.685-1.000)] or post-operative CFs with RFs [AUC (95%CI) = 0.894 (0.757-1.000)] also showed good discrimination.

Conclusion

The radiomics-based HMCAS might be a promising tool to predict the prognoses of patients with AIS after MT.

A CT-based radiomics nomogram for classification of intraparenchymal hyperdense areas in patients with acute ischemic stroke following mechanical thrombectomy treatment

Objectives

To develop and validate a radiomic-based model for differentiating hemorrhage from iodinated contrast extravasation of intraparenchymal hyperdense areas (HDA) following mechanical thrombectomy treatment in acute ischemic stroke.

Methods

A total of 100 and four patients with intraparenchymal HDA on initial post-operative CT were included in this study. The patients who met criteria were divided into a primary and a validation cohort. A training cohort was constructed using Synthetic Minority Oversampling Technique on the primary cohort to achieve group balance. Thereafter, a radiomics score was calculated and the radiomic model was constructed. Clinical factors were assessed to build clinical model. Combined with the Rad-score and independent clinical factors, a combined model was constructed. Different models were assessed using the area under the receiver operator characteristic curves. The combined model was visualized as nomogram, and assessed with calibration and clinical usefulness.

Results

Cardiogenic diseases, intraoperative tirofiban administration and preoperative national institute of health stroke scale were selected as independent predictors to construct the clinical model with area under curve (AUC) of 0.756 and 0.693 in the training and validation cohort, respectively. Our data demonstrated that the radiomic model showed good discrimination in the training (AUC, 0.955) and validation cohort (AUC, 0.869). The combined nomogram model showed optimal discrimination in the training (AUC, 0.972) and validation cohort (AUC, 0.926). Decision curve analysis demonstrated the combined model had a higher overall net benefit in differentiating hemorrhage from iodinated contrast extravasation in terms of clinical usefulness.

Conclusions

The nomogram shows favorable efficacy for differentiating hemorrhage from iodinated contrast extravasation, which might provide an individualized tool for precision therapy.

Radiomics model using preoperative computed tomography angiography images to differentiate new from old emboli of acute lower limb arterial embolism

Our purpose was to devise a radiomics model using preoperative computed tomography angiography (CTA) images to differentiate new from old emboli of acute lower limb arterial embolism. 57 patients (95 regions of interest; training set: n = 57; internal validation set: n = 38) with femoral popliteal acute lower limb arterial embolism confirmed by pathology and with preoperative CTA images were retrospectively analyzed. We selected the best prediction model according to the model performance tested by area under the curve (AUC) analysis across 1,000 iterations of prediction from three most common machine learning methods: support vector machine, feed-forward neural network (FNN), and random forest, through several steps of feature selection. Then, the selected best model was also validated in an external validation dataset (n = 24). The established radiomics signature had good predictive efficacy. FNN exhibited the best model performance on the training and validation groups: its AUC value was 0.960 (95% CI, 0.899-1). The accuracy of this model was 89.5%, and its sensitivity and specificity were 0.938 and 0.864, respectively. The AUC of external validation dataset was 0.793. Our radiomics model based on preoperative CTA images is valuable. The radiomics approach of preoperative CTA to differentiate new emboli from old is feasible.

Radiomics of CTA is feasible in identifying muscle ischemia

BACKGROUND

Advanced models based on computed tomography angiography (CTA) radiomics features in discriminating muscle ischemia from normal condition are lacking.

PURPOSE

To investigate the feasibility of radiomics of CTA in discriminating ischemic muscle from normal muscle.

MATERIAL AND METHODS

A total of 102 patients (51 ischemia and 51 non-ischemia) were analyzed using a CTA radiomics method. The radiomics features of muscle were compared between ischemic and normal cases. The maximum relevance minimum redundancy (mRMR) algorithm and least absolute shrinkage and selection operator (LASSO) logistic regression model were used. The receiver operating characteristic (ROC) curve was used to determine the performance of radiomics signature.

RESULTS

Thirty-nine CTA radiomics features were significantly different between the two groups (P < 0.05). By LASSO, six features were used to construct a model. The signature area under the curve was 0.92 and 0.91 in the training and validation cohorts, respectively. The sensitivity and specificity of the signature were 92% and 86% for the training cohort, and 80% and 94% for the validation cohort, respectively.

CONCLUSION

CTA radiomics signature is useful in identifying ischemic muscle in selected patients.

Outcome Prediction Based on Automatically Extracted Infarct Core Image Features in Patients with Acute Ischemic Stroke

Infarct volume (FIV) on follow-up diffusion-weighted imaging (FU-DWI) is only moderately associated with functional outcome in acute ischemic stroke patients. However, FU-DWI may contain other imaging biomarkers that could aid in improving outcome prediction models for acute ischemic stroke. We included FU-DWI data from the HERMES, ISLES, and MR CLEAN-NO IV databases. Lesions were segmented using a deep learning model trained on the HERMES and ISLES datasets. We assessed the performance of three classifiers in predicting functional independence for the MR CLEAN-NO IV trial cohort based on: (1) FIV alone, (2) the most important features obtained from a trained convolutional autoencoder (CAE), and (3) radiomics. Furthermore, we investigated feature importance in the radiomic-feature-based model. For outcome prediction, we included 206 patients: 144 scans were included in the training set, 21 in the validation set, and 41 in the test set. The classifiers that included the CAE and the radiomic features showed AUC values of 0.88 and 0.81, respectively, while the model based on FIV had an AUC of 0.79. This difference was not found to be statistically significant. Feature importance results showed that lesion intensity heterogeneity received more weight than lesion volume in outcome prediction. This study suggests that predictions of functional outcome should not be based on FIV alone and that FU-DWI images capture additional prognostic information.

Nomograms predict prognosis and hospitalization time using non-contrast CT and CT perfusion in patients with ischemic stroke

Background

Stroke is a major disease with high morbidity and mortality worldwide. Currently, there is no quantitative method to evaluate the short-term prognosis and length of hospitalization of patients.

Purpose

We aimed to develop nomograms as prognosis predictors based on imaging characteristics from non-contrast computed tomography (NCCT) and CT perfusion (CTP) and clinical characteristics for predicting activity of daily living (ADL) and hospitalization time of patients with ischemic stroke.

Materials and methods

A total of 476 patients were enrolled in the study and divided into the training set (n = 381) and testing set (n = 95). Each of them owned NCCT and CTP images. We propose to extract imaging features representing as the Alberta stroke program early CT score (ASPECTS) values from NCCT, ischemic lesion volumes from CBF, and TMAX maps from CTP. Based on imaging features and clinical characteristics, we addressed two main issues: (1) predicting prognosis according to the Barthel index (BI)–binary logistic regression analysis was employed for feature selection, and the resulting nomogram was assessed in terms of discrimination capability, calibration, and clinical utility and (2) predicting the hospitalization time of patients–the Cox proportional hazard model was used for this purpose. After feature selection, another specific nomogram was established with calibration curves and time-dependent ROC curves for evaluation.

Results

In the task of predicting binary prognosis outcome, a nomogram was constructed with the area under the curve (AUC) value of 0.883 (95% CI: 0.781–0.985), the accuracy of 0.853, and F1-scores of 0.909 in the testing set. We further tried to predict discharge BI into four classes. Similar performance was achieved as an AUC of 0.890 in the testing set. In the task of predicting hospitalization time, the Cox proportional hazard model was used. The concordance index of the model was 0.700 (SE = 0.019), and AUCs for predicting discharge at a specific week were higher than 0.80, which demonstrated the superior performance of the model.

Conclusion

The novel non-invasive NCCT- and CTP-based nomograms could predict short-term ADL and hospitalization time of patients with ischemic stroke, thus allowing a personalized clinical outcome prediction and showing great potential in improving clinical efficiency.

Summary

Combining NCCT- and CTP-based nomograms could accurately predict short-term outcomes of patients with ischemic stroke, including whose discharge BI and the length of hospital stay.

Key Results

Using a large dataset of 1,310 patients, we show a novel nomogram with a good performance in predicting discharge BI class of patients (AUCs > 0.850). The second nomogram owns an excellent ability to predict the length of hospital stay (AUCs > 0.800).

Feasibility of a Clinical-Radiomics Model to Predict the Outcomes of Acute Ischemic Stroke

Objective

To develop a model incorporating radiomic features and clinical factors to accurately predict acute ischemic stroke (AIS) outcomes.

Materials and Methods

Data from 522 AIS patients (382 male [73.2%]; mean age ± standard deviation, 58.9 ± 11.5 years) were randomly divided into the training (n = 311) and validation cohorts (n = 211). According to the modified Rankin Scale (mRS) at 6 months after hospital discharge, prognosis was dichotomized into good (mRS ≤ 2) and poor (mRS > 2); 1310 radiomics features were extracted from diffusion-weighted imaging and apparent diffusion coefficient maps. The minimum redundancy maximum relevance algorithm and the least absolute shrinkage and selection operator logistic regression method were implemented to select the features and establish a radiomics model. Univariable and multivariable logistic regression analyses were performed to identify the clinical factors and construct a clinical model. Ultimately, a multivariable logistic regression analysis incorporating independent clinical factors and radiomics score was implemented to establish the final combined prediction model using a backward step-down selection procedure, and a clinical-radiomics nomogram was developed. The models were evaluated using calibration, receiver operating characteristic (ROC), and decision curve analyses.

Results

Age, sex, stroke history, diabetes, baseline mRS, baseline National Institutes of Health Stroke Scale score, and radiomics score were independent predictors of AIS outcomes. The area under the ROC curve of the clinical-radiomics model was 0.868 (95% confidence interval, 0.825–0.910) in the training cohort and 0.890 (0.844–0.936) in the validation cohort, which was significantly larger than that of the clinical or radiomics models. The clinical radiomics nomogram was well calibrated (p > 0.05). The decision curve analysis indicated its clinical usefulness.

Conclusion

The clinical-radiomics model outperformed individual clinical or radiomics models and achieved satisfactory performance in predicting AIS outcomes.

Fully Automated Thrombus Segmentation on CT Images of Patients with Acute Ischemic Stroke

Thrombus imaging characteristics are associated with treatment success and functional outcomes in stroke patients. However, assessing these characteristics based on manual annotations is labor intensive and subject to observer bias. Therefore, we aimed to create an automated pipeline for consistent and fast full thrombus segmentation. We used multi-center, multi-scanner datasets of anterior circulation stroke patients with baseline NCCT and CTA for training (n = 228) and testing (n = 100). We first found the occlusion location using StrokeViewer LVO and created a bounding box around it. Subsequently, we trained dual modality U-Net based convolutional neural networks (CNNs) to segment the thrombus inside this bounding box. We experimented with: (1) U-Net with two input channels for NCCT and CTA, and U-Nets with two encoders where (2) concatenate, (3) add, and (4) weighted-sum operators were used for feature fusion. Furthermore, we proposed a dynamic bounding box algorithm to adjust the bounding box. The dynamic bounding box algorithm reduces the missed cases but does not improve Dice. The two-encoder U-Net with a weighted-sum feature fusion shows the best performance (surface Dice 0.78, Dice 0.62, and 4% missed cases). Final segmentation results have high spatial accuracies and can therefore be used to determine thrombus characteristics and potentially benefit radiologists in clinical practice.

Deep Learning-Enabled Clinically Applicable CT Planbox for Stroke With High Accuracy and Repeatability

Background

Computed tomography (CT) plays an essential role in classifying stroke, quantifying penumbra size and supporting stroke-relevant radiomics studies. However, it is difficult to acquire standard, accurate and repeatable images during follow-up. Therefore, we invented an intelligent CT to evaluate stroke during the entire follow-up.

Methods

We deployed a region proposal network (RPN) and V-Net to endow traditional CT with intelligence. Specifically, facial detection was accomplished by identifying adjacent jaw positions through training and testing an RPN on 76,382 human faces using a preinstalled 2-dimensional camera; two regions of interest (ROIs) were segmented by V-Net on another training set with 295 subjects, and the moving distance of scanning couch was calculated based on a pre-generated calibration table. Multiple cohorts including 1,124 patients were used for performance validation under three clinical scenarios.

Results

Cranial Automatic Planbox Imaging Towards AmeLiorating neuroscience (CAPITAL)-CT was invented. RPN model had an error distance of 4.46 ± 0.02 pixels with a success rate of 98.7% in the training set and 100% with 2.23 ± 0.10 pixels in the testing set. V-Net-derived segmentation maintained a clinically tolerable distance error, within 3 mm on average, and all lines presented with a tolerable angle error, within 3° on average in all boundaries. Real-time, accurate, and repeatable automatic scanning was accomplished with and a lower radiation exposure dose (all P < 0.001).

Conclusions

CAPITAL-CT generated standard and reproducible images that could simplify the work of radiologists, which would be of great help in the follow-up of stroke patients and in multifield research in neuroscience.

A Machine Learning Model Predicts the Outcome of SRS for Residual Arteriovenous malformations after partial embolization- A Real-World Clinical Obstacle

PURPOSE

To propose a machine learning (ML) model predicting the favorable outcome of stereotactic radiosurgery (SRS) for residual brain arteriovenous malformation (bAVM) after partial embolization.

MATERIALS AND METHODS

One hundred and thirty bAVM patients who underwent partial embolization followed by SRS were retrospectively reviewed. Patients were randomly split into training datasets (n=100) and testing datasets (n=30). Radiomics and dosimetric features were extracted from pre-SRS treatment images. Feature selection was performed to select appropriate radiomics and dosimetric features. Three ML algorithms were applied to construct models using selected features respectively. A total of 9 models were trained to predict favorable outcomes (obliteration without complication) of bAVMs. The efficacy of these models was evaluated on the testing dataset using mean accuracy (ACC) and area under the receiver operating characteristic curve (AUROC).

RESULTS

The obliteration rate of this cohort was 70.77% (92/130) with a mean follow-up period of 43.8 (Range 12-108 months) months. Favorable outcomes were achieved in 89 (68.46%) patients. Four radiomics features and 7 dosimetric features were selected for ML model construction. The dosimetric SVM showed the best performance on the training dataset, with an ACC and AUC of 0.74 and 0.78 respectively. The dosimetric SVM model also showed the best performance on the testing dataset where the ACC and AUC were 0.83 and 0.77 respectively.

CONCLUSION

Dosimetric features are good predictors of prognosis for patients with partially embolized bAVM followed by SRS therapy. The use of ML models is an innovative method for predicting favorable outcomes of partially embolized bAVM followed by SRS therapy.

Emerging Applications of Radiomics in Neurological Disorders: A Review

Radiomics has achieved significant momentum in radiology research and can reveal image information invisible to radiologists' eyes. Radiomics first evolved for oncologic imaging. Oncologic applications (histopathology, tumor grading, gene mutation analysis, patient survival, and treatment response prediction) of radiomics are widespread. However, it is not limited to oncologic analysis, and any digital medical images can benefit from radiomics analysis. This article reviews the current literature on radiomics in non-oncologic, neurological disorders including ischemic strokes, hemorrhagic stroke, cerebral aneurysms, and demyelinating disorders.

Radiographic Characteristics of Mild Ischemic Stroke Patients With Visible Intracranial Occlusion: The INTERRSeCT Study

BACKGROUND AND PURPOSE

Early neurological deterioration occurs in one-third of mild strokes primarily due to the presence of a relevant intracranial occlusion. We studied vascular occlusive patterns, thrombus characteristics, and recanalization rates in these patients.

METHODS

Among patients enrolled in INTERRSeCT (Identifying New Approaches to Optimize Thrombus Characterization for Predicting Early Recanalization and Reperfusion With IV Alteplase and Other Treatments Using Serial CT Angiography), a multicenter prospective study of acute ischemic strokes with a visible intracranial occlusion, we compared characteristics of mild (National Institutes of Health Stroke Scale score, ≤5) to moderate/severe strokes.

RESULTS

Among 575 patients, 12.9% had a National Institutes of Health Stroke Scale score ≤5 (median age, 70.5 [63-79]; 58% male; median National Institutes of Health Stroke Scale score, 4 [2-4]). Demographics and vascular risk factors were similar between the two groups. As compared with those with a National Institutes of Health Stroke Scale score >5, mild patients had longer symptom onset to assessment times (onset to computed tomography [240 versus 167 minutes] and computed tomography angiography [246 versus 172 minutes]), more distal occlusions (M3, anterior cerebral artery and posterior cerebral artery; 22% versus 6%), higher clot burden score (median, 9 [6-9] versus 6 [4-9]), similar favorable thrombus permeability (residual flow grades I-II, 21% versus 19%), higher collateral flow (9.1 versus 7.6), and lower intravenous alteplase treatment rates (55% versus 85%). Mild patients were more likely to recanalize (revised arterial occlusion scale score 2b/3, 45%; 49% with alteplase) compared with moderate/severe strokes (26%; 29% with alteplase). In an adjusted model for sex, alteplase, residual flow, and time between the two vessel imagings, intravenous alteplase use (odds ratio, 3.80 [95% CI, 1.11-13.00]) and residual flow grade (odds ratio, 8.70 [95% CI, 1.26-60.13]) were associated with successful recanalization among mild patients.

CONCLUSIONS

Mild strokes with visible intracranial occlusions have different vascular occlusive patterns but similar thrombus permeability compared with moderate/severe strokes. Higher thrombus permeability and alteplase use were associated with successful recanalization, although the majority do not recanalize. Randomized controlled trials are needed to assess the efficacy of new thrombolytics and endovascular therapy in this population.

Identifying Thrombus on Non-Contrast CT in Patients with Acute Ischemic Stroke

The hyperdense sign is a marker of thrombus in non-contrast computed tomography (NCCT) datasets. The aim of this work was to determine optimal Hounsfield unit (HU) thresholds for thrombus segmentation in thin-slice non-contrast CT (NCCT) and use these thresholds to generate 3D thrombus models. Patients with thin-slice baseline NCCT (≤2.5 mm) and MCA-M1 occlusions were included. CTA was registered to NCCT, and three regions of interest (ROIs) were placed in the NCCT, including: the thrombus, contralateral brain tissue, and contralateral patent MCA-M1 artery. Optimal HU thresholds differentiating the thrombus from non-thrombus tissue voxels were calculated using receiver operating characteristic analysis. Linear regression analysis was used to predict the optimal HU threshold for discriminating the clot only based on the average contralateral vessel HU or contralateral parenchyma HU. Three-dimensional models from 70 participants using standard (45 HU) and patient-specific thresholds were generated and compared to CTA clot characteristics. The optimal HU threshold discriminating thrombus in NCCT from other structures varied with a median of 51 (IQR: 49-55). Experts chose 3D models derived using patient-specific HU models as corresponding better to the thrombus seen in CTA in 83.8% (31/37) of cases. Patient-specific HU thresholds for segmenting the thrombus in NCCT can be derived using normal parenchyma. Thrombus segmentation using patient-specific HU thresholds is superior to conventional 45 HU thresholds.

FLAIR and ADC Image-Based Radiomics Features as Predictive Biomarkers of Unfavorable Outcome in Patients With Acute Ischemic Stroke

At present, it is still challenging to predict the clinical outcome of acute ischemic stroke (AIS). In this retrospective study, we explored whether radiomics features extracted from fluid-attenuated inversion recovery (FLAIR) and apparent diffusion coefficient (ADC) images can predict clinical outcome of patients with AIS. Patients with AIS were divided into a training (n = 110) and an external validation (n = 80) sets. A total of 753 radiomics features were extracted from each FLAIR and ADC image of the 190 patients. Interquartile range (IQR), Wilcoxon rank sum test, and least absolute shrinkage and selection operator (LASSO) were used to reduce the feature dimension. The six strongest radiomics features were related to an unfavorable outcome of AIS. A logistic regression analysis was employed for selection of potential predominating clinical and conventional magnetic resonance imaging (MRI) factors. Subsequently, we developed several models based on clinical and conventional MRI factors and radiomics features to predict the outcome of AIS patients. For predicting unfavorable outcome [modified Rankin scale (mRS) > 2] in the training set, the area under the receiver operating characteristic curve (AUC) of ADC radiomics model was 0.772, FLAIR radiomics model 0.731, ADC and FLAIR radiomics model 0.815, clinical model 0.791, and clinical and conventional MRI model 0.782. In the external validation set, the AUCs for the prediction with ADC radiomics model was 0.792, FLAIR radiomics model 0.707, ADC and FLAIR radiomics model 0.825, clinical model 0.763, and clinical and conventional MRI model 0.751. When adding radiomics features to the combined model, the AUCs for predicting unfavorable outcome in the training and external validation sets were 0.926 and 0.864, respectively. Our results indicate that the radiomics features extracted from FLAIR and ADC can be instrumental biomarkers to predict unfavorable clinical outcome of AIS and would additionally improve predictive performance when adding to combined model.

Radiology-Pathology Correlations of Intracranial Clots: Current Theories, Clinical Applications, and Future Directions

In recent years, there has been substantial progression in the field of stroke clot/thrombus imaging. Thrombus imaging aims to deduce the histologic composition of the clot through evaluation of various imaging characteristics. If the histology of a thrombus can be reliably determined by noninvasive imaging methods, critical information may be extrapolated about its expected response to treatment and about the patient's clinical outcome. Crucially, as we move into an era of stroke therapy individualization, determination of the histologic composition of a clot may be able to guide precise and targeted therapeutic effort. Most radiologists, however, remain largely unfamiliar with the topic of clot imaging. This article will review the current literature regarding clot imaging, including its histologic backdrop, the correlation of images with cellular components and treatment responsiveness, and future expectations.

Deep Semantic Segmentation Feature-Based Radiomics for the Classification Tasks in Medical Image Analysis

Recently, an emerging trend in medical image classification is to combine radiomics framework with deep learning classification network in an integrated system. Although this combination is efficient in some tasks, the deep learning-based classification network is often difficult to capture an effective representation of lesion regions, and prone to face the challenge of overfitting, leading to unreliable features and inaccurate results, especially when the sizes of the lesions are small or the training dataset is small. In addition, these combinations mostly lack an effective feature selection mechanism, which makes it difficult to obtain the optimal feature selection. In this paper, we introduce a novel and effective deep semantic segmentation feature-based radiomics (DSFR) framework to overcome the above-mentioned challenges, which consists of two modules: the deep semantic feature extraction module and the feature selection module. Specifically, the extraction module is utilized to extract hierarchical semantic features of the lesions from a trained segmentation network. The feature selection module aims to select the most representative features by using a novel feature similarity adaptation algorithm. Experiments are extensively conducted to evaluate our method in two clinical tasks: the pathological grading prediction in pancreatic neuroendocrine neoplasms (pNENs), and the prediction of thrombolytic therapy efficacy in deep venous thrombosis (DVT). Experimental results on both tasks demonstrate that the proposed method consistently outperforms the state-of-the-art approaches by a large margin.