Brain Tumor Segmentation for Multi-Modal MRI with Missing Information.

Deep convolutional neural networks (DCNNs) have shown promise in brain tumor segmentation from multi-modal MRI sequences, accommodating heterogeneity in tumor shape and appearance. The fusion of multiple MRI sequences allows networks to explore complementary tumor information for segmentation. However, developing a network that maintains clinical relevance in situations where certain MRI sequence(s) might be unavailable or unusual poses a significant challenge. While one solution is to train multiple models with different MRI sequence combinations, it is impractical to train every model from all possible sequence combinations. In this paper, we propose a DCNN-based brain tumor segmentation framework incorporating a novel sequence dropout technique in which networks are trained to be robust to missing MRI sequences while employing all other available sequences. Experiments were performed on the RSNA-ASNR-MICCAI BraTS 2021 Challenge dataset. When all MRI sequences were available, there were no significant differences in performance of the model with and without dropout for enhanced tumor (ET), tumor (TC), and whole tumor (WT) (p-values 1.000, 1.000, 0.799, respectively), demonstrating that the addition of dropout improves robustness without hindering overall performance. When key sequences were unavailable, the network with sequence dropout performed significantly better. For example, when tested on only T1, T2, and FLAIR sequences together, DSC for ET, TC, and WT increased from 0.143 to 0.486, 0.431 to 0.680, and 0.854 to 0.901, respectively. Sequence dropout represents a relatively simple yet effective approach for brain tumor segmentation with missing MRI sequences.

Machine Learning Models for Predicting Cycloplegic Refractive Error and Myopia Status Based on Non-Cycloplegic Data in Chinese Students.

Purpose: To develop and validate machine learning (ML) models for predicting cycloplegic refractive error and myopia status using noncycloplegic refractive error and biometric data. Methods: Cross-sectional study of children aged five to 18 years who underwent biometry and autorefraction before and after cycloplegia. Myopia was defined as cycloplegic spherical equivalent refraction (SER) ≤-0.5 Diopter (D). Models were evaluated for predicting SER using R2 and mean absolute error (MAE) and myopia status using area under the receiver operating characteristic (ROC) curve (AUC). Best-performing models were further evaluated using sensitivity/specificity and comparison of observed versus predicted myopia prevalence rate overall and in each age group. Independent data sets were used for training (n = 1938) and validation (n = 1476). Results: In the validation dataset, ML models predicted cycloplegic SER with high R2 (0.913-0.935) and low MAE (0.393-0.480 D). The AUC for predicting myopia was high (0.984-0.987). The best-performing model for SER (XGBoost) had high sensitivity and specificity (91.1% and 97.2%). Random forest (RF), the best-performing model for myopia, had high sensitivity and specificity (92.2% and 96.9%). Within each age group, difference between predicted and actual myopia prevalence was within 4%. Conclusions: Using noncycloplegic refractive error and ocular biometric data, ML models performed well for predicting cycloplegic SER and myopia status. When measuring cycloplegic SER is not feasible, ML may provide a useful tool for estimating cycloplegic SER and myopia prevalence rate in epidemiological studies. Translational Relevance: Using ML to predict cycloplegic refraction based on noncycloplegic data is a powerful tool for large, population-based studies of refractive error.

Active Learning in Brain Tumor Segmentation with Uncertainty Sampling and Annotation Redundancy Restriction.

Deep learning models have demonstrated great potential in medical imaging but are limited by the expensive, large volume of annotations required. To address this, we compared different active learning strategies by training models on subsets of the most informative images using real-world clinical datasets for brain tumor segmentation and proposing a framework that minimizes the data needed while maintaining performance. Then, 638 multi-institutional brain tumor magnetic resonance imaging scans were used to train three-dimensional U-net models and compare active learning strategies. Uncertainty estimation techniques including Bayesian estimation with dropout, bootstrapping, and margins sampling were compared to random query. Strategies to avoid annotating similar images were also considered. We determined the minimum data necessary to achieve performance equivalent to the model trained on the full dataset (α = 0.05). Bayesian approximation with dropout at training and testing showed results equivalent to that of the full data model (target) with around 30% of the training data needed by random query to achieve target performance (p = 0.018). Annotation redundancy restriction techniques can reduce the training data needed by random query to achieve target performance by 20%. We investigated various active learning strategies to minimize the annotation burden for three-dimensional brain tumor segmentation. Dropout uncertainty estimation achieved target performance with the least annotated data.

Evaluation of Multiple Machine Learning Models for Predicting Number of Anti-VEGF Injections in the Comparison of AMD Treatment Trials (CATT).

Purpose: To apply machine learning models for predicting the number of pro re nata (PRN) injections of antivascular endothelial growth factor (anti-VEGF) for neovascular age-related macular degeneration (nAMD) in two years in the Comparison of AMD (age-related macular degeneration) Treatments Trials. Methods: The data from 493 eligible participants randomized to PRN treatment of ranibizumab or bevacizumab were used for training (n = 393) machine learning models including support-vector machine (SVM), random forest, and extreme gradient boosting (XGBoost) models. Model performances of prediction using clinical and image data from baseline, weeks 4, 8, and 12 were evaluated by the area under the receiver operating characteristic curve (AUC) for predicting few (≤8) or many (≥19) injections, by R2 and mean absolute error (MAE) for predicting the total number of injections in two years. The best model was selected for final validation on a test dataset (n = 100). Results: Using training data up to week 12, the models achieved AUCs of 0.79-0.82 and 0.79-0.81 for predicting few and many injections, respectively, with R2 of 0.34-0.36 (MAE = 4.45-4.58 injections) for predicting total injections in two years from cross-validation. In final validation on the test dataset, the SVM model had AUCs of 0.77 and 0.82 for predicting few and many injections, respectively, with R2 of 0.44 (MAE = 3.92 injections). Important features included fluid in optical coherence tomography, lesion characteristics, and treatment trajectory in the first three months. Conclusions: Machine learning models using loading dose phase data have the potential to predict two-year anti-VEGF demand for nAMD and quantify feature importance for these predictions. Translational Relevance: Prediction of anti-VEGF injections using machine learning models from readily available data, after further validation on independent datasets, has the potential to help optimize treatment protocols and outcomes for nAMD patients in an individualized manner.

Predicting Visual Acuity Responses to Anti-VEGF Treatment in the Comparison of Age-related Macular Degeneration Treatments Trials Using Machine Learning.

PURPOSE: To evaluate multiple machine learning (ML) models for predicting 2-year visual acuity (VA) responses to anti-vascular endothelial growth factor (anti-VEGF) treatment in the Comparison of Age-related Macular Degeneration (AMD) Treatments Trials (CATT) for patients with neovascular AMD (nAMD). DESIGN: Secondary analysis of public data from a randomized clinical trial. PARTICIPANTS: A total of 1029 CATT participants who completed 2 years of follow-up with untreated active nAMD and baseline VA between 20/25 and 20/320 in the study eye. METHODS: Five ML models (support vector machine, random forest, extreme gradient boosting, multilayer perceptron neural network, and lasso) were applied to clinical and image data from baseline and weeks 4, 8, and 12 for predicting 4 VA outcomes (≥ 15-letter VA gain, ≥ 15-letter VA loss, VA change from baseline, and actual VA) at 2 years. The CATT data from 1029 participants were randomly split for training (n = 717), from which the models were trained using 10-fold cross-validation, and for final validation on a test data set (n = 312). MAIN OUTCOME MEASURES: Performances of ML models were assessed by R2 and mean absolute error (MAE) for predicting VA change from baseline and actual VA at 2 years, by the area under the receiver operating characteristic curve (AUC) for predicting ≥ 15-letter VA gain and loss from baseline. RESULTS: Using training data up to week 12, the ML models from cross-validation achieved mean R2 of 0.24 to 0.29 (MAE = 9.1-9.8 letters) for predicting VA change and 0.37 to 0.41 (MAE = 9.3-10.2 letters) for predicting actual VA at 2 years. The mean AUCs for predicting ≥ 15-letter VA gain and loss at 2 years was 0.84 to 0.85 and 0.58 to 0.73, respectively. In final validation on the test data set up to week 12, the models had an R2 of 0.33 to 0.38 (MAE = 8.9-9.9 letters) for predicting VA change, an R2 of 0.37 to 0.45 (MAE = 8.8-10.2 letters) for predicting actual VA at 2 years, and AUCs of 0.85 to 0.87 and 0.67 to 0.79 for predicting ≥ 15-letter VA gain and loss, respectively. CONCLUSIONS: Machine learning models have the potential to predict 2-year VA response to anti-VEGF treatment using clinical and imaging features from the loading dose phase, which can aid in decision-making around treatment protocols for patients with nAMD. FINANCIAL DISCLOSURE(S): The author(s) have no proprietary or commercial interest in any materials discussed in this article.

CYP2D6 Substrate Dispensing Among Patients Dispensed Mirabegron: An Administrative Claims Analysis.

BACKGROUND: Overactive bladder (OAB) is characterized by the presence of bothersome urinary symptoms. Pharmacologic treatment options for OAB include anticholinergics and ß3-adrenergic agonists. Use of ß3-adrenergic agonists may result in similar treatment efficacy with a decreased side effect profile compared with anticholinergics because high anticholinergic burden is associated with cardiovascular and neurologic side effects. However, the ß3-adrenergic agonist mirabegron, one of two approved drugs within this class, is a moderate cytochrome P450 (CYP) 2D6 inhibitor, and coadministration of drugs that are CYP2D6 substrates with mirabegron may lead to adverse drug effects. OBJECTIVE: The aim of this study was to quantify how often CYP2D6 substrates were dispensed in patients receiving mirabegron among adults of any age and among those ≥ 65 years of age. METHODS: In this retrospective descriptive analysis, a deidentified administrative claims database in the United States, IQVIA PharMetrics® Plus, was used to identify dispensing claims for CYP2D6 substrates and mirabegron from November 2012 to September 2019. Prevalence of CYP2D6 substrate dispensing was assessed in patients dispensed mirabegron among all adults ≥ 18 years old and additionally among a cohort of those ≥ 65 years old. Patient baseline profiles at the time of mirabegron and CYP2D6 substrate codispensing and at the time of mirabegron dispensing were compared. CYP2D6 substrates were categorized as those with the potential for increased risk of QT prolongation, with anticholinergic properties, with narrow therapeutic index (NTI), contraindicated or having a black box warning when used with CYP2D6 inhibitors, or used for depression or other psychiatric disease. Dispensing data and patient profiles were summarized descriptively. RESULTS: Overall, 68.5% of adults ≥ 18 years old dispensed mirabegron had overlapping dispensings for one or more CYP2D6 substrate; 60.6% and 53.6% had overlapping dispensings for CYP2D6 substrates with anticholinergic properties or risk of QT prolongation, respectively. CYP2D6 substrates with NTI, contraindicated with CYP2D6 inhibitors, or for psychiatric use were codispensed in 17.7%, 16.6%, and 38.0% of adult mirabegron users, respectively. Mirabegron users receiving one or more concurrent CYP2D6 substrate were more likely to be older, have more comorbidities and baseline polypharmacy, and have increased healthcare resource utilization compared with those without concurrent CYP2D6 substrates. Commonly codispensed CYP2D6 substrates included hydrocodone, oxycodone, tramadol, metoprolol, and tamsulosin. Findings were similar for patients in the older cohort (≥ 65 years old), with 72.1% receiving overlapping CYP2D6 substrates. CONCLUSIONS: Codispensing of CYP2D6 substrates, especially those with anticholinergic properties or risk of QT prolongation, was common among adults and older adults receiving mirabegron. Results highlight the need for improved awareness of CYP2D6 substrate prescribing among patients receiving pharmacologic treatment for OAB that inhibits the CYP2D6 pathway.

Translatability Analysis of National Institutes of Health-Funded Biomedical Research That Applies Artificial Intelligence.

Importance: Despite the rapid growth of interest and diversity in applications of artificial intelligence (AI) to biomedical research, there are limited objective ways to characterize the potential for use of AI in clinical practice. Objective: To examine what types of medical AI have the greatest estimated translational impact (ie, ability to lead to development that has measurable value for human health) potential. Design, Setting, and Participants: In this cohort study, research grants related to AI awarded between January 1, 1985, and December 31, 2020, were identified from a National Institutes of Health (NIH) award database. The text content for each award was entered into a Natural Language Processing (NLP) clustering algorithm. An NIH database was also used to extract citation data, including the number of citations and approximate potential to translate (APT) score for published articles associated with the granted awards to create proxies for translatability. Exposures: Unsupervised assignment of AI-related research awards to application topics using NLP. Main Outcomes and Measures: Annualized citations per $1 million funding (ACOF) and average APT score for award-associated articles, grouped by application topic. The APT score is a machine-learning based metric created by the NIH Office of Portfolio Analysis that quantifies the likelihood of future citation by a clinical article. Results: A total of 16 629 NIH awards related to AI were included in the analysis, and 75 applications of AI were identified. Total annual funding for AI grew from $17.4 million in 1985 to $1.43 billion in 2020. By average APT, interpersonal communication technologies (0.488; 95% CI, 0.472-0.504) and population genetics (0.463; 95% CI, 0.453-0.472) had the highest translatability; environmental health (ACOF, 1038) and applications focused on the electronic health record (ACOF, 489) also had high translatability. The category of applications related to biochemical analysis was found to have low translatability by both metrics (average APT, 0.393; 95% CI, 0.388-0.398; ACOF, 246). Conclusions and Relevance: Based on this study's findings, data on grants from the NIH can apparently be used to identify and characterize medical applications of AI to understand changes in academic productivity, funding support, and potential for translational impact. This method may be extended to characterize other research domains.