A machine learning approach for predicting perihematomal edema expansion in patients with intracerebral hemorrhage

An interpretable artificial intelligence model based on CT for prognosis of intracerebral hemorrhage: a multicenter study

OBJECTIVES

To develop and validate a novel interpretable artificial intelligence (AI) model that integrates radiomic features, deep learning features, and imaging features at multiple semantic levels to predict the prognosis of intracerebral hemorrhage (ICH) patients at 6 months post-onset.

MATERIALS AND METHODS

Retrospectively enrolled 222 patients with ICH for Non-contrast Computed Tomography (NCCT) images and clinical data, who were divided into a training cohort (n = 186, medical center 1) and an external testing cohort (n = 36, medical center 2). Following image preprocessing, the entire hematoma region was segmented by two radiologists as the volume of interest (VOI). Pyradiomics algorithm library was utilized to extract 1762 radiomics features, while a deep convolutional neural network (EfficientnetV2-L) was employed to extract 1000 deep learning features. Additionally, radiologists evaluated imaging features. Based on the three different modalities of features mentioned above, the Random Forest (RF) model was trained, resulting in three models (Radiomics Model, Radiomics-Clinical Model, and DL-Radiomics-Clinical Model). The performance and clinical utility of the models were assessed using the Area Under the Receiver Operating Characteristic Curve (AUC), calibration curve, and Decision Curve Analysis (DCA), with AUC compared using the DeLong test. Furthermore, this study employs three methods, Shapley Additive Explanations (SHAP), Grad-CAM, and Guided Grad-CAM, to conduct a multidimensional interpretability analysis of model decisions.

RESULTS

The Radiomics-Clinical Model and DL-Radiomics-Clinical Model exhibited relatively good predictive performance, with an AUC of 0.86 [95% Confidence Intervals (CI): 0.71, 0.95; P < 0.01] and 0.89 (95% CI: 0.74, 0.97; P < 0.01), respectively, in the external testing cohort.

CONCLUSION

The multimodal explainable AI model proposed in this study can accurately predict the prognosis of ICH. Interpretability methods such as SHAP, Grad-CAM, and Guided Grad-Cam partially address the interpretability limitations of AI models. Integrating multimodal imaging features can effectively improve the performance of the model.

CLINICAL RELEVANCE STATEMENT

Predicting the prognosis of patients with ICH is a key objective in emergency care. Accurate and efficient prognostic tools can effectively prevent, manage, and monitor adverse events in ICH patients, maximizing treatment outcomes.

Predicting Outcome of Patients With Cerebral Hemorrhage Using a Computed Tomography-Based Interpretable Radiomics Model: A Multicenter Study

OBJECTIVE

The aim of this study was to develop and validate an interpretable and highly generalizable multimodal radiomics model for predicting the prognosis of patients with cerebral hemorrhage.

METHODS

This retrospective study involved 237 patients with cerebral hemorrhage from 3 medical centers, of which a training cohort of 186 patients (medical center 1) was selected and 51 patients from medical center 2 and medical center 3 were used as an external testing cohort. A total of 1762 radiomics features were extracted from nonenhanced computed tomography using Pyradiomics, and the relevant macroscopic imaging features and clinical factors were evaluated by 2 experienced radiologists. A radiomics model was established based on radiomics features using the random forest algorithm, and a radiomics-clinical model was further trained by combining radiomics features, clinical factors, and macroscopic imaging features. The performance of the models was evaluated using area under the curve (AUC), sensitivity, specificity, and calibration curves. Additionally, a novel SHAP (SHAPley Additive exPlanations) method was used to provide quantitative interpretability analysis for the optimal model.

RESULTS

The radiomics-clinical model demonstrated superior predictive performance overall, with an AUC of 0.88 (95% confidence interval, 0.76-0.95; P < 0.01). Compared with the radiomics model (AUC, 0.85; 95% confidence interval, 0.72-0.94; P < 0.01), there was a 0.03 improvement in AUC. Furthermore, SHAP analysis revealed that the fusion features, rad score and clinical rad score, made significant contributions to the model's decision-making process.

CONCLUSION

Both proposed prognostic models for cerebral hemorrhage demonstrated high predictive levels, and the addition of macroscopic imaging features effectively improved the prognostic ability of the radiomics-clinical model. The radiomics-clinical model provides a higher level of predictive performance and model decision-making basis for the risk prognosis of cerebral hemorrhage.

Artificial Intelligence and Machine Learning Applications in Critically Ill Brain Injured Patients

The utilization of Artificial Intelligence (AI) and Machine Learning (ML) is paving the way for significant strides in patient diagnosis, treatment, and prognostication in neurocritical care. These technologies offer the potential to unravel complex patterns within vast datasets ranging from vast clinical data and EEG (electroencephalogram) readings to advanced cerebral imaging facilitating a more nuanced understanding of patient conditions. Despite their promise, the implementation of AI and ML faces substantial hurdles. Historical biases within training data, the challenge of interpreting multifaceted data streams, and the "black box" nature of ML algorithms present barriers to widespread clinical adoption. Moreover, ethical considerations around data privacy and the need for transparent, explainable models remain paramount to ensure trust and efficacy in clinical decision-making.This article reflects on the emergence of AI and ML as integral tools in neurocritical care, discussing their roles from the perspective of both their scientific promise and the associated challenges. We underscore the importance of extensive validation in diverse clinical settings to ensure the generalizability of ML models, particularly considering their potential to inform critical medical decisions such as withdrawal of life-sustaining therapies. Advancement in computational capabilities is essential for implementing ML in clinical settings, allowing for real-time analysis and decision support at the point of care. As AI and ML are poised to become commonplace in clinical practice, it is incumbent upon health care professionals to understand and oversee these technologies, ensuring they adhere to the highest safety standards and contribute to the realization of personalized medicine. This engagement will be pivotal in integrating AI and ML into patient care, optimizing outcomes in neurocritical care through informed and data-driven decision-making.

Knowledge structure and global trends of machine learning in stroke over the past decade: A scientometric analysis

Objective

Machine learning (ML) models have been widely applied in stroke prediction, diagnosis, treatment, and prognosis assessment. We aimed to conduct a comprehensive scientometrics analysis of studies related to ML in stroke and reveal its current status, knowledge structure, and global trends.

Methods

All documents related to ML in stroke were retrieved from the Web of Science database on March 15, 2023. We refined the documents by including only original articles and reviews in the English language. The literature published over the past decade was imported into scientometrics software for influence detection and collaborative network analysis.

Results

2389 related publications were included. The annual publication outputs demonstrated explosive growth, with an average growth rate of 63.99 %. Among the 90 countries/regions involved, the United States (729 articles) and China (636 articles) were the most productive countries. Frontiers in Neurology was the most prolific journal with 94 articles. 234 highly cited articles, each with more than 31 citations, were detected. Keyword analysis revealed a total of 5333 keywords, with a predominant focus on the application of ML models in the early diagnosis, classification, and prediction of "acute ischemic stroke" and "atrial fibrillation-related stroke". The keyword "classification" had the first and longest burst, spanning from 2013 to 2018. 'Upport vector machine' got the strongest burst strength with 6.2. Keywords such as 'mechanical thrombectomy', 'expression', and 'prognosis' experienced bursts in 2022 and have continued to be prominent.

Conclusion

The applications of ML in stroke are increasingly diverse and extensive, with researchers showing growing interest over the past decade. However, the clinical application of ML in stroke is still in its early stages, and several limitations and challenges need to be addressed for its widespread adoption in clinical practice.

A Radiomics Model Based on CT Images Combined with Multiple Machine Learning Models to Predict the Prognosis of Spontaneous Intracerebral Hemorrhage

OBJECTIVE

We aimed to construct 3 predictive models, including a clinical model, a radiomics model, and a combined model, to forecast the discharge prognosis of patients with intracerebral hemorrhage on admission.

METHODS

A retrospective study was conducted, involving a total of 161 patients with intracerebral hemorrhage (ICH). At a ratio of 7:3, 115 of these patients were assigned to the training cohort, and 46 of these patients were assigned to the validation cohort. To produce the radionics signature and pick the features to use in its construction, the least absolute shrinkage and selection operator (LASSO) regression was applied. Five machine models were applied, and the optimal model was chosen to construct the radionics model. A clinical model was constructed using univariate and stepwise analysis to identify independent risk variables for poor outcomes at discharge. A predictive combined model nomogram was generated by integrating the clinical model and radiomics model. The performance of the nomogram was assessed in the training cohort and validated in the validation cohort. Analyses of the receiver operating characteristic curve (ROC), the calibration curve, and the decision curve were performed to assess the performance of the combined model.

RESULTS

This study encompassed a cohort of 161 individuals diagnosed with intracerebral hemorrhage (ICH), consisting of 110 males and 51 females. Utilizing the modified Rankin Scale (mRS) at discharge, the analysis revealed that 89 patients (55.3%) had a good prognosis, while 72 patients (44.7%) had a poor prognosis. Only 8 out of 1130 radiomics features were selected and used in conjunction with the LR algorithm to develop the radiomics model. Sex, IVH, GCS score, and ICH volume were determined to be independent predictors of poor outcomes at the time of discharge. The AUC values of the combined model, radiomics model, and clinical model were 0.8583, 0.8364, and 0.7579 in the training cohort, and 0.9153, 0.8692, and 0.7114 in the validation cohort, respectively. The combined model nomogram exhibited good calibration and clinical benefit in both the training and validation cohorts. The decision curve analysis (DCA) displays that the combined model obtained the highest net benefit compared to the radiomics model and clinics model in the training cohort.

CONCLUSIONS

The combined model demonstrates reliability and efficacy in predicting the discharge prognosis of ICH, enabling physicians to perform individualized risk assessments, and make optimal choices about patients with ICH.

Automated cerebral hemorrhage volume calculation and stability detection using automated software

Introduction

The measurement of intracerebral hemorrhage (ICH) volume is important for management, particularly in evaluating expansion on subsequent imaging. However manual volumetric analysis is time-consuming, especially in busy hospital settings. We aimed to use automated Rapid Hyperdensity software to accurately measure ICH volume across repeated imaging.

Methods

We identified ICH cases, with repeat imaging conducted within 24 hours, from two randomized clinical trials where enrollment was not based on ICH volume. Scans were excluded if there was (1) severe CT artifacts, (2) prior neurosurgical procedures, (3) recent intravenous contrast, or (4) ICH < 1 ml. Manual ICH measurements were conducted by one neuroimaging expert using MIPAV software and compared to the performance of automated software.

Results

127 patients were included with median baseline ICH volume manually measured at 18.18 cc (IQR: 7.31 - 35.71) compared to automated detection of 18.93 cc (IQR: 7.55, 37.88). The two modalities were highly correlated (r = 0.994, p < 0.001). On repeat imaging, the median absolute difference in ICH volume was 0.68cc (IQR: -0.60-4.87) compared to automated detection at 0.68cc (IQR: -0.45-4.63). These absolute differences were also highly correlated (r = 0.941, p < 0.001), with the ability of the automated software to detect ICH expansion with a Sensitivity of 94.12% and Specificity 97.27%.

Conclusion

In our proof-of-concept study, the automated software has high reliability in its ability to quickly determine IPH volume with high sensitivity and specificity and to detect expansion on subsequent imaging.