Predicting Hematoma Expansion Research Articles

Improving the Robustness of Deep-Learning Models in Predicting Hematoma Expansion from Admission Head CT.

Robustness against input data perturbations is essential for deploying deep-learning models in clinical practice. Adversarial attacks involve subtle, voxel-level manipulations of scans to increase deep-learning models' prediction errors. Testing deep-learning model performance on examples of adversarial images provides a measure of robustness, and including adversarial images in the training set can improve the model's robustness. In this study, we examined adversarial training and input modifications to improve the robustness of deep-learning models in predicting hematoma expansion (HE) from admission head CTs of patients with acute intracerebral hemorrhage (ICH). We used a multicenter cohort of n=890 patients for cross-validation/training, and a cohort of n=684 consecutive ICH patients from two stroke centers for independent validation. Fast Gradient Sign Method (FGSM) and Projected Gradient Descent (PGD) adversarial attacks were applied for training and testing. We developed and tested four different models to predict ≥3mL, ≥6mL, ≥9mL, and ≥12mL HE in independent validation cohort applying Receiver Operating Characteristics (ROC) Area Under the Curve (AUC). We examined varying mixtures of adversarial and non-perturbed (clean) scans for training as well as including additional input from the hyperparameter-free Otsu multi-threshold segmentation for model. When deep-learning models trained solely on clean scans were tested with PGD and FGSM adversarial images, the average HE prediction AUC dropped from 0.8 to 0.67 and 0.71, respectively. Overall, the best performing strategy to improve model robustness was training with 5-to-3 mix of clean and PGD adversarial scans and addition of Otsu multi-threshold segmentation to model input, increasing the average AUC to 0.77 against both PGD and FGSM adversarial attacks. Adversarial training with FGSM improved robustness against similar type attack but offered limited cross-attack robustness against PGD-type images. Adversarial training and inclusion of threshold-based segmentation as an additional input can improve deep-learning model robustness in prediction of HE from admission head CTs in acute ICH. ATACH-2= Antihypertensive Treatment of Acute Cerebral Hemorrhage; AUC= Area Under the Curve; Dice=Dice coefficient; CNN= Convolutional Neural Network; FGSM= Fast Gradient Sign Method; ICH= Intracerebral hemorrhage; HD= Hausdorff distance; HE= Hematoma expansion; PGD= Projected Gradient Descent; ROC= Receiver Operating Characteristics; VS= Volume similarity.

Reliability of ABC/2 volumetric estimation in spontaneous intracerebral hemorrhage for hematoma expansion prediction scores.

Prediction scores for hematoma expansion in spontaneous intracerebral hemorrhage (ICH), such as the 9-point and BRAIN scores, were developed predominantly using planimetry to measure hematoma volume. In this study, we aim to investigate whether the ABC/2 formula, which is known to overestimate hematoma volume, can be reliably used as a substitute for planimetry in these prediction scores. A total of 429 patients from four hospitals were retrospectively enrolled. CT scan and clinical data at admission and follow-up CT scan were collected. The 9-point and BRAIN scores were calculated using hematoma volume from ABC/2 and planimetry. Hematoma expansion was assessed using hematoma volume from planimetry. The median hematoma volume measured by ABC/2 was 11.97 ml (interquartile range [IQR], 4.8-30.0), whereas the volume measured by planimetry was 11.70 ml (IQR, 4.9-26.6). The median measurement error between ABC/2 and planimetry was 0.30 ml (IQR, -0.72-2.87). ABC/2 overestimated hematoma volume in 244 patients (56.9%) compared to planimetry. In the 9-point score, the area under the curves (AUCs) for predicting hematoma expansion were 0.735 (95% confidence interval [CI], 0.675-0.796) with ABC/2 and 0.732 (95% CI, 0.672-0.793) with planimetry. In the BRAIN score, the AUCs were 0.753 (95% CI, 0.693-0.813) with ABC/2 and 0.745 (95% CI, 0.688-0.803) with planimetry. The 9-point and BRAIN scores using hematoma volume measured by ABC/2 and planimetry showed good performance in predicting hematoma expansion in ICH. ABC/2 volumetric estimation proved to be reliable for these scores.

Black hole sign under anticoagulant therapy: A retrospective comparison of warfarin and direct oral anticoagulants.

Direct oral anticoagulants (DOAC) have rapidly replaced warfarin. Intracerebral hemorrhage (ICH) is known to be one of the most severe side effects of anticoagulant drugs. The black hole (BH) sign is reportedly a valid radiological sign for predicting hematoma expansion in acute ICH. Here, we hypothesized that the frequency of BH signs might differ between warfarin and DOAC treatment, and critically evaluated the clinical value of the BH sign in acute ICH under warfarin versus DOAC therapy. Patients with acute ICH under anticoagulant therapy were enrolled. Hematoma volumes were measured by ABC/2. Radiologists blinded to the clinical information determined the presence or absence of the BH sign on CT images. This study defined a more than 12.5 ml increase in hematoma volume as cases with "expanded hematoma". We analyzed 111 acute ICH patients under anticoagulant therapy. Among them, 21 patients were treated with antagonists in this cohort. Multivariate logistic regression analysis revealed that the presence of ventricular perforation (p=0.02; adjusted odds ratio (OR): 3.51, 95% confidence interval (CI): 1.32 - 10.2) and the BH sign (p<0.01; adjusted OR: 4.86, 95% CI: 1.73 - 14.3) were significantly different between expanded and non-expanded hematoma cases. Comparison of hematoma volume and the presence of the BH sign between warfarin and DOAC cases indicated significant differences in maximum hematoma volume (p=0.03) and presence of the BH sign (p<0.01). The increase in hematoma volume was significantly greater when the BH sign was present under warfarin therapy (p=0.05). In contrast, the increase in hematoma volume did not differ between cases with and without the BH sign in patients under DOAC therapy (p=0.14) CONCLUSIONS: The BH sign is a useful radiological signature to predict the expansion of acute ICH under anticoagulant therapy. ICH under warfarin tended to present the BH sign more frequently than that under DOAC. The results also showed that the BH sign is more reliable under warfarin than under DOAC therapy in ICH patients. AF = atrial fibrillation; BH = black hole; DOAC = direct oral anticoagulants; HU = Hounsfield Unit; ICH= intracerebral hemorrhage.

Morphological characteristics of CT blend sign predict hematoma expansion and outcomes in intracerebral hemorrhage in elderly patients.

The underlying basis of the blend sign on brain computed tomography (CT) in patients with intracerebral hemorrhage (ICH) is unclear. Few studies have examined the morphological alterations in the CT blend sign in ICH. Therefore, we assessed changes in the CT blend sign and their association with hematoma expansion (HE) and adverse outcomes in ICH patients. We recorded the clinical and radiographic parameters of patients with ICH and blend sign on brain CT. The patients were categorized into two groups, with changes in the relatively hypoattenuating region of the blend sign (CHB group) and with no changes in the relatively hypoattenuating region of the blend sign (NHB groups). We performed univariate and multivariate logistic regression analyses to examine the correlations between CHB and HE and poor outcomes. Furthermore, receiver operating characteristic curve analysis was used to confirm the predictive power of CHB. In total, 183 patients were included in the study, of whom 74 (40.4%) demonstrated changes in the hypoattenuating region of the blend sign, whereas 109 (59.6%) did not. Compared with the NHB group, patients in the CHB group exhibited significantly higher levels of HE and adverse outcomes. After adjustment for confounding factors, CHB was independently associated with HE (odds ratio, 19.401 [95% CI, 7.217-52.163]; p < 0.001) and poor 3-month outcomes (odds ratio, 2.638 [95% CI, 1.391-5.003]; p = 0.003) in the multivariate analysis. The sensitivity, specificity, positive predictive value, and negative predictive value of CHB for predicting HE were 0.877, 0.768, 0.791, and 0.862, respectively, whereas these values for predicting poor outcomes were 0.789, 0.641, 0.688, and 0.752, respectively. Changes of a hypoattenuating region within the blend sign have good predictive accuracy for HE and short-term adverse outcomes in elderly patients with ICH. ClinicalTrials.gov, NCT05548530.

Development and validation of the nomogram model derived non-contrast CT score to predict hematoma expansion in patients with spontaneous intracerebral hemorrhage

Develop and validate new non-contrast computed tomography (NCCT) score to predict hematoma expansion (HE) in spontaneous intracerebral hemorrhage (SICH) patients based on hematoma's shape irregularity and density heterogeneity. Retrospective study was conducted among 136 patients for development and 90 patients for validation at two separate hospitals. SICH patients with NCCT scanned within 6 hours of symptoms and follow-up NCCT scanned within 24 hours were enrolled. Black hole sign and blend sign were integrated as combined heterogeneity; likewise, satellite sign and island sign were integrated as combined irregularity. Binary logistic regression analysis screened the covariates associated with HE. Nomogram was generated using the predicted value of binary logistic regression model to derive NCCT score to predict HE. A total of 65 patients had HE in developmental cohort, where history of hypertension [odds ratio (OR) 2.56; 95% CI 1.169-5.607; P=0.019], initial NCCT time ≤ 3 hours (OR 2.50; 95% CI 1.169-5.327; P=0.018), combined heterogeneity (OR 2.50; 95% CI 1.160-5.365; P=0.019), and combined irregularity (OR 2.63; 95% CI 1.164-5.942; P=0.020) were independently associated with HE. A score was derived and a single point was allocated to each independently associated variable. HE was observed in 35 patients in validation cohort, which showed a proportional increase in the probability of HE with an increase in score accumulated. New four-point NCCT score to predict HE was developed and validated, which may be regarded as fair predictive score where advance facilities are rarely available.

Hybrid clinical-radiomics model based on fully automatic segmentation for predicting the early expansion of spontaneous intracerebral hemorrhage: A multi-center study

BackgroundEarly prediction of hematoma expansion (HE) is important for the development of therapeutic strategies for spontaneous intracerebral hemorrhage (sICH). Radiomics can help to predict early hematoma expansion in intracerebral hemorrhage. However, complex image processing procedures, especially hematoma segmentation, are time-consuming and dependent on assessor experience. We provide a fully automated hematoma segmentation method, and construct a hybrid predictive model for risk stratification of hematoma expansion. PurposeTo propose an automatic approach for predicting early hemorrhage expansion after spontaneous intracerebral hemorrhage using deep-learning and radiomics methods. MethodsA total of 258 patients with sICH were retrospectively enrolled for model construction and internal validation, while another two cohorts (n=87, 149) were employed for independent validation. For hemorrhage segmentation, an iterative segmentation procedure was performed to delineate the area using an nnU-Net framework. Radiomics models of intra-hemorrhage and multiscale peri-hemorrhage were established and evaluated, and the best discriminative-scale peri-hemorrhage radiomics model was selected for further analysis. Combining clinical factors and intra- and peri-hemorrhage radiomics signatures, a hybrid nomogram was constructed for the early HE prediction using multivariate logistic regression. For model validation, the receiver operating characteristic (ROC) curve analyses and DeLong test were used to evaluate the performances of the constructed models, and the calibration curve and decision curve analysis were performed for clinical application. ResultsOur iterative auto-segmentation model showed satisfactory results for hematoma segmentation in all four cohorts. The Dice similarity coefficient of this hematoma segmentation model reached 0.90, showing an expert-level accuracy in hematoma segmentation. The consumed time of the efficient delineation was significantly decreased, from 18 min to less than 2 min, with the assistance of the auto-segmentation model. The radiomics model of 2-mm peri-hemorrhage had a preferable area under ROC curve (AUC) of 0.840 (95 % confidence interval [CI]: 0.768, 0.912) compared with the original (0-mm dilatation) model with an AUC of 0.796 (95 % CI: 0.717, 0.875). The clinical–radiomics hybrid model showed better performances for HE prediction, with AUC of 0.853, 0.852, 0.772, and 0.818 in the training, internal validation, and independent validation cohorts 1 and 2, respectively. ConclusionsThe fully automatic clinical–radiomics model based on deep learning and radiomics exhibits a good ability for hematoma segmentation and a favorable performance in stratifying HE risks.

Systematic Evaluation of Hematoma Expansion Models in Spontaneous Intracerebral Hemorrhage: A Meta-Analysis and Meta-Regression Approach.

Accurate prediction of hematoma expansion (HE) in spontaneous intracerebral hemorrhage (sICH) is crucial for tailoring patient-specific treatments and improving outcomes. Recent advancements have yielded numerous HE risk factors and predictive models. This study aims to evaluate the characteristics and efficacy of existing HE prediction models, offering insights for performance enhancement. A comprehensive search was conducted in PubMed for observational studies and randomized controlled trials focusing on HE prediction, written in English. The prediction models were categorized based on their incorporated features and modeling methodology. Rigorous quality and bias assessments were performed. A meta-analysis of studies reporting C-statistics was executed to assess and compare the performance of current HE prediction models. Meta-regression was utilized to explore heterogeneity sources. From 358 initial records, 22 studies were deemed eligible, encompassing traditional models, hematoma imaging feature models, and models based on artificial intelligence or radiomics. Meta-analysis of 11 studies, involving 12,087 sICH patients, revealed an aggregated C-statistic of 0.74 (95% CI: 0.69-0.78) across seven HE prediction models. Eight characteristics related to development cohorts were identified as key factors contributing to performance variability among these models. The findings indicate that the current predictive capacity for HE risk remains suboptimal. Enhanced accuracy in HE prediction is vital for effectively targeting patient populations most likely to benefit from tailored treatment strategies.

Prediction of hematoma expansion in spontaneous intracerebral hemorrhage using a multimodal neural network

Hematoma expansion occasionally occurs in patients with intracerebral hemorrhage (ICH), associating with poor outcome. Multimodal neural networks incorporating convolutional neural network (CNN) analysis of images and neural network analysis of tabular data are known to show promising results in prediction and classification tasks. We aimed to develop a reliable multimodal neural network model that comprehensively analyzes CT images and clinical variables to predict hematoma expansion. We retrospectively enrolled ICH patients at four hospitals between 2017 and 2021, assigning patients from three hospitals to the training and validation dataset and patients from one hospital to the test dataset. Admission CT images and clinical variables were collected. CT findings were evaluated by experts. Three types of models were developed and trained: (1) a CNN model analyzing CT images, (2) a multimodal CNN model analyzing CT images and clinical variables, and (3) a non-CNN model analyzing CT findings and clinical variables with machine learning. The models were evaluated on the test dataset, focusing first on sensitivity and second on area under the receiver operating curve (AUC). Two hundred seventy-three patients (median age, 71 years [59–79]; 159 men) in the training and validation dataset and 106 patients (median age, 70 years [62–82]; 63 men) in the test dataset were included. Sensitivity and AUC of a CNN model were 1.000 (95% confidence interval [CI] 0.768–1.000) and 0.755 (95% CI 0.704–0.807); those of a multimodal CNN model were 1.000 (95% CI 0.768–1.000) and 0.799 (95% CI 0.749–0.849); and those of a non-CNN model were 0.857 (95% CI 0.572–0.982) and 0.733 (95% CI 0.625–0.840). We developed a multimodal neural network model incorporating CNN analysis of CT images and neural network analysis of clinical variables to predict hematoma expansion in ICH. The model was externally validated and showed the best performance of all the models.

CT radiomics combined with clinical and radiological factors predict hematoma expansion in hypertensive intracerebral hemorrhage

ObjectivesThis study aimed to establish a hematoma expansion (HE) prediction model for hypertensive intracerebral hemorrhage (HICH) patients by combining CT radiomics, clinical information, and conventional imaging signs.MethodsA retrospective continuous collection of HICH patients from three medical centers was divided into a training set (n = 555), a validation set (n = 239), and a test set (n = 77). Extract radiomics features from baseline CT plain scan images and combine them with clinical information and conventional imaging signs to construct radiomics models, clinical imaging sign models, and hybrid models, respectively. The models will be evaluated using the area under the curve (AUC), clinical decision curve analysis (DCA), net reclassification index (NRI), and integrated discrimination improvement (IDI).ResultsIn the training, validation, and testing sets, the radiomics model predicts an AUC of HE of 0.885, 0.827, and 0.894, respectively, while the clinical imaging sign model predicts an AUC of HE of 0.759, 0.725, and 0.765, respectively. Glasgow coma scale score at admission, first CT hematoma volume, irregular hematoma shape, and radiomics score were used to construct a hybrid model, with AUCs of 0.901, 0.838, and 0.917, respectively. The DCA shows that the hybrid model had the highest net profit rate. Compared with the radiomics model and the clinical imaging sign model, the hybrid model showed an increase in NRI and IDI.ConclusionThe hybrid model based on CT radiomics combined with clinical and radiological factors can effectively individualize the evaluation of the risk of HE in patients with HICH.Clinical relevance statementCT radiomics combined with clinical information and conventional imaging signs can identify HICH patients with a high risk of HE and provide a basis for clinical-targeted treatment.Key PointsHE is an important prognostic factor in patients with HICH.The hybrid model predicted HE with training, validation, and test AUCs of 0.901, 0.838, and 0.917, respectively.This model provides a tool for a personalized clinical assessment of early HE risk.

Leakage Sign Is a Reliable Predictor of Hematoma Expansion in Acute Epidural Hematoma

Hematoma expansion (H-Ex) in small-/medium-sized acute epidural hematoma (AEDH) cases upon emergency admission is critical. Predicting H-Ex can lead to early surgical interventions, improving outcomes, and eliminating the need to check for expansion via computed tomography (CT). This study aimed to identify the most reliable predictors of AEDH expansion. We retrospectively collected data from patients with pure AEDH not requiring surgical treatment upon emergency admission from 2012 to 2022. We assessed clinical and laboratory data, time from injury to the first CT, and time to follow-up CT. Factors predictive of H-Ex on the second follow-up CT, including the leakage sign (LS), were analyzed. A total of 23 patients with pure AEDH without surgery at admission were included, and LS was positive in 18. Thirteen patients showed H-Ex. The H-Ex group showed a significantly higher rate of positive LS and a lower mean platelet count than the group without H-Ex. LS's predictive value for AEDH expansion showed 100% sensitivity and 50% specificity. All patients with negative LS and normal platelet counts showed no H-Ex. Analyzing the time from injury to the first CT suggested that LS (+) within 120minutes strongly predicted H-Ex. Reconstructed three-dimensional images of the leakage point on the skull revealed multiple mottled bleeding points on the dural surface. LS can predict H-Ex in patients with pure AEDH for whom emergency surgery is unnecessary at admission. The time from injury and platelet counts must also be considered.

Hematoma expansion prediction based on SMOTE and XGBoost algorithm

Hematoma expansion (HE) is a high risky symptom with high rate of occurrence for patients who have undergone spontaneous intracerebral hemorrhage (ICH) after a major accident or illness. Correct prediction of the occurrence of HE in advance is critical to help the doctors to determine the next step medical treatment. Most existing studies focus only on the occurrence of HE within 6 h after the occurrence of ICH, while in reality a considerable number of patients have HE after the first 6 h but within 24 h. In this study, based on the medical doctors recommendation, we focus on prediction of the occurrence of HE within 24 h, as well as the occurrence of HE every 6 h within 24 h. Based on the demographics and computer tomography (CT) image extraction information, we used the XGBoost method to predict the occurrence of HE within 24 h. In this study, to solve the issue of highly imbalanced data set, which is a frequent case in medical data analysis, we used the SMOTE algorithm for data augmentation. To evaluate our method, we used a data set consisting of 582 patients records, and compared the results of proposed method as well as few machine learning methods. Our experiments show that XGBoost achieved the best prediction performance on the balanced dataset processed by the SMOTE algorithm with an accuracy of 0.82 and F1-score of 0.82. Moreover, our proposed method predicts the occurrence of HE within 6, 12, 18 and 24 h at the accuracy of 0.89, 0.82, 0.87 and 0.94, indicating that the HE occurrence within 24 h can be predicted accurately by the proposed method.

Hematoma expansion prediction in patients with intracerebral hemorrhage using a deep learning approach

Hematoma expansion prediction in patients with intracerebral hemorrhage using a deep learning approach

HE-Mind: A model for automatically predicting hematoma expansion after spontaneous intracerebral hemorrhage

PurposeTo develop and validate an end-to-end model for automatically predicting hematoma expansion (HE) after spontaneous intracerebral hemorrhage (sICH) using a novel deep learning framework. MethodsThis multicenter retrospective study collected cranial noncontrast computed tomography (NCCT) images of 490 patients with sICH at admission for model training (n = 236), internal testing (n = 60), and external testing (n = 194). A HE-Mind model was designed to predict HE, which consists of a densely connected U-net for segmentation process, a multi-instance learning strategy for resolving label ambiguity and a Siamese network for classification process. Two radiomics models based on support vector machine or logistic regression and two deep learning models based on residual network or Swin transformer were developed for performance comparison. Reader experiments including physician diagnosis mode and artificial intelligence mode were conducted for efficiency comparison. ResultsThe HE-Mind model showed better performance compared to the comparative models in predicting HE, with areas under the curve of 0.849 and 0.809 in the internal and external test sets respectively. With the assistance of the HE-Mind model, the predictive accuracy and work efficiency of the emergency physician, junior radiologist, and senior radiologist were significantly improved, with accuracies of 0.768, 0.789, and 0.809 respectively, and reporting times of 7.26 s, 5.08 s, and 3.99 s respectively. ConclusionsThe HE-Mind model could rapidly and automatically process the NCCT data and predict HE after sICH within three seconds, indicating its potential to assist physicians in the clinical diagnosis workflow of HE.

Red cell distribution width to lymphocyte ratio could serve as a new inflammatory biomarker for predicting hematoma expansion in patients with intracerebral hemorrhage

BackgroundHematoma expansion is a critical factor associated with increased mortality and adverse outcomes in patients with intracerebral hemorrhage (ICH). Identifying and preventing hematoma expansion early on is crucial for effective therapeutic intervention. This study aimed to investigate the potential association between the Red cell distribution width to lymphocyte ratio (RDWLR) and hematoma expansion in ICH patients.MethodsWe conducted a retrospective analysis of clinical data from 303 ICH patients treated at our department between May 2018 and May 2023. Demographic, clinical, radiological, and laboratory data, including RDWLR upon admission, were assessed. Binary logistic regression analysis was employed to determine independent associations between various variables and hematoma expansion.ResultsThe study included 303 ICH patients, comprising 167 (55.1%) males and 136 (44.9%) females, with a mean age of 65.25 ± 7.32 years at admission. Hematoma expansion occurred in 73 (24.1%) cases. Multivariate analysis revealed correlations between hematoma volume at baseline (OR, 2.73; 95% CI: 1.45 -4,78; P < 0.001), admission systolic blood pressure (OR, 2.98 ; 95% CI: 1.54–4.98; P < 0.001), Glasgow Coma Scale (GCS) (OR, 1.58; 95% CI: 1.25–2.46; P = 0.017), and RDWLR (OR, 1.58; 95% CI: 1.13–2.85; P = 0.022) and hematoma expansion in these patients.ConclusionsOur findings suggest that RDWLR could serve as a new inflammatory biomarker for hematoma expansion in ICH patients. This cost-effective and readily available biomarker has the potential for early prediction of hematoma expansion in these patients.

Predicting hematoma expansion using machine learning: An exploratory analysis of the ATACH 2 trial

IntroductionHematoma expansion (HE) in patients with intracerebral hemorrhage (ICH) is a key predictor of poor prognosis and potentially amenable to treatment. This study aimed to build a classification model to predict HE in patients with ICH using deep learning algorithms without using advanced radiological features. MethodsData from the ATACH-2 trial (Antihypertensive Treatment of Acute Cerebral Hemorrhage) was utilized. Variables included in the models were chosen as per literature consensus on salient variables associated with HE. HE was defined as increase in either >33% or 6 mL in hematoma volume in the first 24 h. Multiple machine learning algorithms were employed using iterative feature selection and outcome balancing methods. 70% of patients were used for training and 30% for internal validation. We compared the ML models to a logistic regression model and calculated AUC, accuracy, sensitivity and specificity for the internal validation models respective models. ResultsAmong 1000 patients included in the ATACH-2 trial, 924 had the complete parameters which were included in the analytical cohort. The median [interquartile range (IQR)] initial hematoma volume was 9.93.mm3 [5.03–18.17] and 25.2% had HE. The best performing model across all feature selection groups and sampling cohorts was using an artificial neural network (ANN) for HE in the testing cohort with AUC 0.702 [95% CI, 0.631–0.774] with 8 hidden layer nodes The traditional logistic regression yielded AUC 0.658 [95% CI, 0.641–0.675]. All other models performed with less accuracy and lower AUC. Initial hematoma volume, time to initial CT head, and initial SBP emerged as most relevant variables across all best performing models. ConclusionWe developed multiple ML algorithms to predict HE with the ANN classifying the best without advanced radiographic features, although the AUC was only modestly better than other models. A larger, more heterogenous dataset is needed to further build and better generalize the models.

Swirl sign score system: a novel and practical tool for predicting hematoma expansion risk after spontaneous intracerebral haemorrhage.

To methodically analyse the swirl sign and construct a scoring system to predict the risk of hematoma expansion (HE) after spontaneous intracerebral haemorrhage (sICH). We analysed 231 of 683 sICH patients with swirl signs on baseline noncontrast CT (NCCT) images. The characteristics of the swirl sign were analysed, including the number, maximum diameter, shape, boundary, minimum CT value of the swirl sign, and the minimum distance from the swirl sign to the edge of the hematoma. In the development cohort, univariate and multivariate analyses were used to identify independent predictors of HE, and logistic regression analysis was used to construct the swirl sign score system. The swirl sign score system was verified in the validation cohort. The number and the minimum CT value of the swirl sign were independent predictors of HE. The swirl sign score system was constructed (2 points for the number of swirl signs >1 and 1 point for the minimum CT value ≤41 Hounsfield units). The area under the curve of the swirl sign score system in predicting HE was 0.773 and 0.770 in the development and validation groups, respectively. The swirl sign score system is an easy-to-use radiological grading scale that requires only baseline NCCT images to effectively identify subjects at high risk of HE. Our newly developed semiquantitative swirl sign score system greatly improves the ability of swirl sign to predict HE.

Application of deep learning and radiomics in the prediction of hematoma expansion in intracerebral hemorrhage: a fully automated hybrid approach

Spontaneous intracerebral hemorrhage (ICH) is the most severe form of stroke. The timely assessment of early hematoma enlargement and its proper treatment are of great significance in curbing the deterioration and improving the prognosis of patients with ICH. This study aimed to develop an automated hybrid approach to predict hematoma expansion in ICH. The transfer learning method was applied to build a hybrid model based on a convolutional neural network (CNN) to predict the expansion of hematoma. The model integrated (1) a CNN for automated hematoma segmentation and (2) a CNN-based classifier for hematoma expansion prediction that incorporated both 2-dimensional images and the radiomics features of the 3-dimensional hematoma shape. The radiomics feature module had the highest area under the receiver operating characteristic curve (AUC) of 0.58, a precision of 0, a recall of 0, and an average precision (AP) of 0.26. The ResNet50 and Inception_v3 modules had AUCs of 0.79 and 0.93, a precision of 0.56 and 0.86, a recall of 0.42 and 0.75, and an AP of 0.51 and 0.85, respectively. Radiomic with Inception_v3 and Radiomic with ResNet50 had AUCs of 0.95 and 0.81, a precision of 0.90 and 0.57, a recall of 0.79 and 0.17, and an AP of 0.87 and 0.69, respectively. A model using deep learning and radiomics was successfully developed. This model can reliably predict the hematoma expansion of ICH with a fully automated process based on non-contrast computed tomography imaging. Furthermore, the radiomics fusion with the Inception_v3 model had the highest accuracy.

P2Y12 inhibitor use predicts hematoma expansion in patients with intracerebral hemorrhage.

Hematoma expansion (HE) predicts disability and death after acute intracerebral hemorrhage (ICH). Aspirin and anticoagulants have been associated with HE. We tested the hypothesis that P2Y12 inhibitors predict subsequent HE in patients. We explored laboratory measures of P2Y12 inhibition and dual antiplatelet therapy with aspirin (DAPT). We prospectively identified patients with ICH. Platelet activity was measured with the VerifyNow-P2Y12 assay. Hematoma volumes for initial and follow-up CTs were calculated using a validated semi-automated technique. HE was defined as the difference between hematoma volumes on the initial and follow-up CT scans. Nonparametric statistics were performed with Kruskal-Wallis H, and correction for multiple comparisons performed with Dunn's test. In 194 patients, 15 (7.7%) were known to take a P2Y12 inhibitor (clopidogrel in all but one). Patients taking a P2Y12 inhibitor had more HE compared to patients not taking a P2Y12 inhibitor (3.5 [1.2-11.9] vs. 0.1 [-0.8-1.4] mL, p = 0.004). Patients taking DAPT experienced the most HE (7.2 [2.6-13.8] vs. 0.0 [-1.0-1.1] mL, p = 0.04). The use of P2Y12 inhibitors was associated with less P2Y12 activity (178 [149-203] vs. 288 [246-319] P2Y12 reaction units, p = 0.005). Patients taking a P2Y12 inhibitor had more HE and less P2Y12 activity. The effect was most pronounced in patients on DAPT, suggesting a synergistic effect of P2Y12 inhibitors and aspirin with respect to HE. Acute reversal of P2Y12 inhibitors in acute ICH requires further study.

Quantitative imaging for predicting hematoma expansion in intracerebral hemorrhage: A multimodel comparison

BackgroundSeveral studies report that radiomics provides additional information for predicting hematoma expansion in intracerebral hemorrhage (ICH). However, the comparison of diagnostic performance of radiomics for predicting revised hematoma expansion (RHE) remains unclear. MethodsThe cohort comprised 312 consecutive patients with ICH. A total of 1106 radiomics features from seven categories were extracted using Python software. Support vector machines achieved the best performance in both the training and validation datasets. Clinical factors models were constructed to predict RHE. Receiver operating characteristic curve analysis was used to assess the abilities of non-contrast computed tomography (NCCT) signs, radiomics features, and combined models to predict RHE. ResultsWe finally selected the top 21 features for predicting RHE. After univariate analysis, 4 clinical factors and 5 NCCT signs were selected for inclusion in the prediction models. In the training and validation dataset, radiomics features had a higher predictive value for RHE (AUC = 0.83) than a single NCCT sign and expansion-prone hematoma. The combined prediction model including radiomics features, clinical factors, and NCCT signs achieved higher predictive performances for RHE (AUC = 0.88) than other combined models. ConclusionsNCCT radiomics features have a good degree of discrimination for predicting RHE in ICH patients. Combined prediction models that include quantitative imaging significantly improve the prediction of RHE, which may assist in the risk stratification of ICH patients for anti-expansion treatments.

An End-to-End Deep Learning Framework for Predicting Hematoma Expansion in Hemorrhagic Stroke Patients from CT Images

Hematoma expansion (HE) occurs in 20% of patients with hemorrhagic stroke within 24 h of onset, and it is associated with a poorer patient outcome. From a clinical point of view, predicting HE from the initial patient computed tomography (CT) image is useful to improve therapeutic decisions and minimize prognosis errors. In this work, we propose an end-to-end deep learning framework for predicting the final hematoma expansion and its corresponding lesion mask. We also explore the problem of having limited data and propose to augment the available dataset with synthetic images. The obtained results show an improved HE prediction when incorporating the use of synthetic images into the model, with a mean Dice score of the HE growth area of 0.506 and an average prediction error in hematoma volume of −3.44 mL. The proposed approach achieved results in line with state-of-the-art methods with far fewer data by using synthetic image generation and without requiring the inclusion of patient clinical data.