Providing clinical context to the spatio-temporal analysis of 4D CT perfusion to predict acute ischemic stroke lesion outcomes.

Acute ischemic stroke is a leading cause of mortality and morbidity worldwide. Timely identification of the extent of a stroke is crucial for effective treatment, whereas spatio-temporal (4D) Computed Tomography Perfusion (CTP) imaging is playing a critical role in this process. Recently, the first deep learning-based methods that leverage the full spatio-temporal nature of perfusion imaging for predicting stroke lesion outcomes have been proposed. However, clinical information is typically not integrated into the learning process, which may be helpful to improve the tissue outcome prediction given the known influence of various factors (i.e., physiological, demographic, and treatment factors) on lesion growth. Cross-attention, a multimodal fusion strategy, has been successfully used to combine information from multiple sources, but it has yet to be applied to stroke lesion outcome prediction. Therefore, this work aimed to develop and evaluate a novel multimodal and spatio-temporal deep learning model that utilizes cross-attention to combine information from 4D CTP and clinical metadata simultaneously to predict stroke lesion outcomes. The proposed model was evaluated using a dataset of 70 acute ischemic stroke patients, demonstrating significantly improved volume estimates (mean error = 19 ml) compared to a baseline unimodal approach (mean error = 35 ml, p< 0.05). The proposed model allows generating attention maps and counterfactual outcome scenarios to investigate the relevance of clinical variables in predicting stroke lesion outcomes at a patient level, helping to provide a better understanding of the model's decision-making process.

Machine learning using multimodal clinical, electroencephalographic, and magnetic resonance imaging data can predict incident depression in adults with epilepsy: A pilot study.

OBJECTIVE: This study was undertaken to develop a multimodal machine learning (ML) approach for predicting incident depression in adults with epilepsy. METHODS: We randomly selected 200 patients from the Calgary Comprehensive Epilepsy Program registry and linked their registry-based clinical data to their first-available clinical electroencephalogram (EEG) and magnetic resonance imaging (MRI) study. We excluded patients with a clinical or Neurological Disorders Depression Inventory for Epilepsy (NDDI-E)-based diagnosis of major depression at baseline. The NDDI-E was used to detect incident depression over a median of 2.4 years of follow-up (interquartile range [IQR] = 1.5-3.3 years). A ReliefF algorithm was applied to clinical as well as quantitative EEG and MRI parameters for feature selection. Six ML algorithms were trained and tested using stratified threefold cross-validation. Multiple metrics were used to assess model performances. RESULTS: Of 200 patients, 150 had EEG and MRI data of sufficient quality for ML, of whom 59 were excluded due to prevalent depression. Therefore, 91 patients (41 women) were included, with a median age of 29 (IQR = 22-44) years. A total of 42 features were selected by ReliefF, none of which was a quantitative MRI or EEG variable. All models had a sensitivity > 80%, and five of six had an F1 score ≥ .72. A multilayer perceptron model had the highest F1 score (median = .74, IQR = .71-.78) and sensitivity (84.3%). Median area under the receiver operating characteristic curve and normalized Matthews correlation coefficient were .70 (IQR = .64-.78) and .57 (IQR = .50-.65), respectively. SIGNIFICANCE: Multimodal ML using baseline features can predict incident depression in this population. Our pilot models demonstrated high accuracy for depression prediction. However, overall performance and calibration can be improved. This model has promise for identifying those at risk for incident depression during follow-up, although efforts to refine it in larger populations along with external validation are required.

Annotation-free prediction of treatment-specific tissue outcome from 4D CT perfusion imaging in acute ischemic stroke.

Acute ischemic stroke is a critical health condition that requires timely intervention. Following admission, clinicians typically use perfusion imaging to facilitate treatment decision-making. While deep learning models leveraging perfusion data have demonstrated the ability to predict post-treatment tissue infarction for individual patients, predictions are often represented as binary or probabilistic masks that are not straightforward to interpret or easy to obtain. Moreover, these models typically rely on large amounts of subjectively segmented data and non-standard perfusion analysis techniques. To address these challenges, we propose a novel deep learning approach that directly predicts follow-up computed tomography images from full spatio-temporal 4D perfusion scans through a temporal compression. The results show that this method leads to realistic follow-up image predictions containing the infarcted tissue outcomes. The proposed compression method achieves comparable prediction results to using perfusion maps as inputs but without the need for perfusion analysis or arterial input function selection. Additionally, separate models trained on 45 patients treated with thrombolysis and 102 treated with thrombectomy showed that each model correctly captured the different patient-specific treatment effects as shown by image difference maps. The findings of this work clearly highlight the potential of our method to provide interpretable stroke treatment decision support without requiring manual annotations.

Exploiting macro- and micro-structural brain changes for improved Parkinson's disease classification from MRI data.

Parkinson's disease (PD) is the second most common neurodegenerative disease. Accurate PD diagnosis is crucial for effective treatment and prognosis but can be challenging, especially at early disease stages. This study aimed to develop and evaluate an explainable deep learning model for PD classification from multimodal neuroimaging data. The model was trained using one of the largest collections of T1-weighted and diffusion-tensor magnetic resonance imaging (MRI) datasets. A total of 1264 datasets from eight different studies were collected, including 611 PD patients and 653 healthy controls (HC). These datasets were pre-processed and non-linearly registered to the MNI PD25 atlas. Six imaging maps describing the macro- and micro-structural integrity of brain tissues complemented with age and sex parameters were used to train a convolutional neural network (CNN) to classify PD/HC subjects. Explainability of the model's decision-making was achieved using SmoothGrad saliency maps, highlighting important brain regions. The CNN was trained using a 75%/10%/15% train/validation/test split stratified by diagnosis, sex, age, and study, achieving a ROC-AUC of 0.89, accuracy of 80.8%, specificity of 82.4%, and sensitivity of 79.1% on the test set. Saliency maps revealed that diffusion tensor imaging data, especially fractional anisotropy, was more important for the classification than T1-weighted data, highlighting subcortical regions such as the brainstem, thalamus, amygdala, hippocampus, and cortical areas. The proposed model, trained on a large multimodal MRI database, can classify PD patients and HC subjects with high accuracy and clinically reasonable explanations, suggesting that micro-structural brain changes play an essential role in the disease course.

Towards objective and systematic evaluation of bias in artificial intelligence for medical imaging.

OBJECTIVE: Artificial intelligence (AI) models trained using medical images for clinical tasks often exhibit bias in the form of subgroup performance disparities. However, since not all sources of bias in real-world medical imaging data are easily identifiable, it is challenging to comprehensively assess their impacts. In this article, we introduce an analysis framework for systematically and objectively investigating the impact of biases in medical images on AI models. MATERIALS AND METHODS: Our framework utilizes synthetic neuroimages with known disease effects and sources of bias. We evaluated the impact of bias effects and the efficacy of 3 bias mitigation strategies in counterfactual data scenarios on a convolutional neural network (CNN) classifier. RESULTS: The analysis revealed that training a CNN model on the datasets containing bias effects resulted in expected subgroup performance disparities. Moreover, reweighing was the most successful bias mitigation strategy for this setup. Finally, we demonstrated that explainable AI methods can aid in investigating the manifestation of bias in the model using this framework. DISCUSSION: The value of this framework is showcased in our findings on the impact of bias scenarios and efficacy of bias mitigation in a deep learning model pipeline. This systematic analysis can be easily expanded to conduct further controlled in silico trials in other investigations of bias in medical imaging AI. CONCLUSION: Our novel methodology for objectively studying bias in medical imaging AI can help support the development of clinical decision-support tools that are robust and responsible.

Risk Factors for Perinatal Arterial Ischemic Stroke: A Machine Learning Approach.

BACKGROUND AND OBJECTIVES: Perinatal arterial ischemic stroke (PAIS) is a focal vascular brain injury presumed to occur between the fetal period and the first 28 days of life. It is the leading cause of hemiparetic cerebral palsy. Multiple maternal, intrapartum, delivery, and fetal factors have been associated with PAIS, but studies are limited by modest sample sizes and complex interactions between factors. Machine learning approaches use large and complex data sets to enable unbiased identification of clinical predictors but have not yet been applied to PAIS. We combined large PAIS data sets and used machine learning methods to identify clinical PAIS factors and compare this data-driven approach with previously described literature-driven clinical prediction models. METHODS: Common data elements from 3 registries with patients with PAIS, the Alberta Perinatal Stroke Project, Canadian Cerebral Palsy Registry, International Pediatric Stroke Study, and a longitudinal cohort of healthy controls (Alberta Pregnancy Outcomes and Nutrition Study), were used to identify potential predictors of PAIS. Inclusion criteria were term birth and idiopathic PAIS (absence of primary causative medical condition). Data including maternal/pregnancy, intrapartum, and neonatal factors were collected between January 2003 and March 2020. Common data elements were entered into a validated random forest machine learning pipeline to identify the highest predictive features and develop a predictive model. Univariable analyses were completed post hoc to assess the relationship between each predictor and outcome. RESULTS: A machine learning model was developed using data from 2,571 neonates, including 527 cases (20%) and 2,044 controls (80%). With a mean of 21 features selected, the random forest machine learning approach predicted the outcome with approximately 86.5% balanced accuracy. Factors that were selected a priori through literature-driven variable selection that were also identified as most important by the machine learning model were maternal age, recreational substance exposure, tobacco exposure, intrapartum maternal fever, and low Apgar score at 5 minutes. Additional variables identified through machine learning included in utero alcohol exposure, infertility, miscarriage, primigravida, meconium, spontaneous vaginal delivery, neonatal head circumference, and 1-minute Apgar score. Overall, the machine learning model performed better (area under the curve [AUC] 0.93) than the literature-driven model (AUC 0.73). DISCUSSION: Machine learning may be an alternative, unbiased method to identify clinical predictors associated with PAIS. Identification of previously suggested and novel clinical factors requires cautious interpretation but supports the multifactorial nature of PAIS pathophysiology. Our results suggest that identification of neonates at risk of PAIS is possible.

Distinct phenotypes of multisystem inflammatory syndrome in children: a cohort study.

BACKGROUND: Multisystem inflammatory syndrome in children (MIS-C) is a severe disease with an unpredictable course and a substantial risk of cardiogenic shock. Our objectives were to (a) compare MIS-C phenotypes across the COVID-19 pandemic, (b) identify features associated with intensive care need and treatment with biologic agents. METHODS: Youth aged 0-18 years, fulfilling the World Health Organization case definition of MIS-C, and admitted to the Alberta Children's Hospital during the first four waves of the COVID-19 pandemic (May 2020-December 2021) were included in this cohort study. Demographic, clinical, biochemical, imaging, and treatment data were captured. RESULTS: Fifty-seven MIS-C patients (median age 6 years, range 0-17) were included. Thirty patients (53%) required intensive care. Patients in the third or fourth wave (indicated as phase 2 of the pandemic) presented with higher peak ferritin (µg/l, median (IQR) = 1134 (409-1806) vs. 370 (249-629), P = 0.001), NT-proBNP (ng/l, median (IQR) = 12,217 (3013-27,161) vs. 3213 (1216-8483), P = 0.02) and D-dimer (mg/l, median (IQR) = 4.81 (2.24-5.37) vs. 2.01 (1.27-3.34), P = 0.004) levels, and higher prevalence of liver enzyme abnormalities (n(%) = 17 (68) vs. 11 (34), P = 0.02), hypoalbuminemia (n(%) = 24 (100) vs. 25 (81), P = 0.03) and thrombocytopenia (n(%) 18 (72) vs. 11 (34), P = 0.007) compared to patients in the first two waves (phase 1). These patients had a higher need of non-invasive/mechanical ventilation (n(%) 4 (16) vs. 0 (0), P = 0.03). Unsupervised clustering analyses classified 47% of the patients in the correct wave and 74% in the correct phase of the pandemic. NT-proBNP was the only significant contributor to the need for intensive care in all applied multivariate regression models. Treatment with biologic agents was significantly associated with peak CRP (mg/l (median, IQR = 240.9 (132.9-319.4) vs. 155.8 (101.0-200.7), P = 0.02) and ferritin levels (µg/l, median (IQR) = 1380 (509-1753) vs. 473 (280-296)). CONCLUSIONS: MIS-C patients in a later stage of the pandemic displayed a more severe phenotype, reflecting the impact of distinct SARS-CoV-2 variants. NT-proBNP emerged as the most crucial feature associated with intensive care need, underscoring the importance of monitoring.

Predicting treatment-specific lesion outcomes in acute ischemic stroke from 4D CT perfusion imaging using spatio-temporal convolutional neural networks.

For the diagnosis and precise treatment of acute ischemic stroke, predicting the final location and volume of lesions is of great clinical interest. Current deep learning-based prediction methods mainly use perfusion parameter maps, which can be calculated from spatio-temporal (4D) CT perfusion (CTP) imaging data, to estimate the tissue outcome of an acute ischemic stroke. However, this calculation relies on a deconvolution operation, an ill-posed problem requiring strong regularization and definition of an arterial input function. Thus, improved predictions might be achievable if the deep learning models were applied directly to acute 4D CTP data rather than perfusion maps. In this work, a novel deep spatio-temporal convolutional neural network is proposed for predicting treatment-dependent stroke lesion outcomes by making full use of raw 4D CTP data. By merging a U-Net-like architecture with temporal convolutional networks, we efficiently process the spatio-temporal information available in CTP datasets to make a tissue outcome prediction. The proposed method was evaluated on 147 patients using a 10-fold cross validation, which demonstrated that the proposed 3D+time model (mean Dice=0.45) significantly outperforms both a 2D+time variant of our approach (mean Dice=0.43) and a state-of-the-art method that uses perfusion maps (mean Dice=0.38). These results show that 4D CTP datasets include more predictive information than perfusion parameter maps, and that the proposed method is an efficient approach to make use of this complex data.

Predicting the tissue outcome of acute ischemic stroke from acute 4D computed tomography perfusion imaging using temporal features and deep learning.

Predicting follow-up lesions from baseline CT perfusion (CTP) datasets in acute ischemic stroke patients is important for clinical decision making. Deep convolutional networks (DCNs) are assumed to be the current state-of-the-art for this task. However, many DCN classifiers have not been validated against the methods currently used in research (random decision forests, RDF) and clinical routine (Tmax thresholding). Specialized DCNs have even been designed to extract complex temporal features directly from spatiotemporal CTP data instead of using standard perfusion parameter maps. However, the benefits of applying deep learning to source or deconvolved CTP data compared to perfusion parameter maps have not been formally investigated so far. In this work, a modular UNet-based DCN is proposed that separates temporal feature extraction from tissue outcome prediction, allowing for both model validation using perfusion parameter maps as well as end-to-end learning from spatiotemporal CTP data. 145 retrospective datasets comprising baseline CTP imaging, perfusion parameter maps, and follow-up non-contrast CT with manual lesion segmentations were assembled from acute ischemic stroke patients treated with intravenous thrombolysis alone (IV; n = 43) or intra-arterial mechanical thrombectomy (IA; n = 102) with or without combined IV. Using the perfusion parameter maps as input, the proposed DCN (mean Dice: 0.287) outperformed the RDF (0.262) and simple Tmax-thresholding (0.249). The performance of the proposed DCN was approximately equal using features optimized from the deconvolved residual curves (0.286) compared to perfusion parameter maps (0.287), while using features optimized from the source concentration-time curves (0.296) provided the best tissue outcome predictions.

Supervised machine learning tools: a tutorial for clinicians.

In an increasingly data-driven world, artificial intelligence is expected to be a key tool for converting big data into tangible benefits and the healthcare domain is no exception to this. Machine learning aims to identify complex patterns in multi-dimensional data and use these uncovered patterns to classify new unseen cases or make data-driven predictions. In recent years, deep neural networks have shown to be capable of producing results that considerably exceed those of conventional machine learning methods for various classification and regression tasks. In this paper, we provide an accessible tutorial of the most important supervised machine learning concepts and methods, including deep learning, which are potentially the most relevant for the medical domain. We aim to take some of the mystery out of machine learning and depict how machine learning models can be useful for medical applications. Finally, this tutorial provides a few practical suggestions for how to properly design a machine learning model for a generic medical problem.