Search for: All records

Abstract AimsNeural network classifiers can detect aortic stenosis (AS) using limited cardiac ultrasound images. While networks perform very well using cart-based imaging, they have never been tested or fine-tuned for use with focused cardiac ultrasound (FoCUS) acquisitions obtained on handheld ultrasound devices. Methods and resultsProspective study performed at Tufts Medical Center. All patients ≥65 years of age referred for clinically indicated transthoracic echocardigraphy (TTE) were eligible for inclusion. Parasternal long axis and parasternal short axis imaging was acquired using a commercially available handheld ultrasound device. Our cart-based AS classifier (trained on ∼10 000 images) was tested on FoCUS imaging from 160 patients. The median age was 74 (inter-quartile range 69–80) years, 50% of patients were women. Thirty patients (18.8%) had some degree of AS. The area under the received operator curve (AUROC) of the cart-based model for detecting AS was 0.87 (95% CI 0.75–0.99) on the FoCUS test set. Last-layer fine-tuning on handheld data established a classifier with AUROC of 0.94 (0.91–0.97). AUROC during temporal external validation was 0.97 (95% CI 0.89–1.0). When performance of the fine-tuned AS classifier was modelled on potential screening environments (2 and 10% AS prevalence), the positive predictive value ranged from 0.72 (0.69–0.76) to 0.88 (0.81–0.97) and negative predictive value ranged from 0.94 (0.94–0.94) to 0.99 (0.99–0.99) respectively. ConclusionOur cart-based machine-learning model for AS showed a drop in performance when tested on handheld ultrasound imaging collected by sonographers. Fine-tuning the AS classifier improved performance and demonstrates potential as a novel approach to detecting AS through automated interpretation of handheld imaging. 

Automated interpretation of ultrasound imaging of the heart (echocardiograms) could improve the detection and treatment of aortic stenosis (AS), a deadly heart disease. However, existing deep learning pipelines for assessing AS from echocardiograms have two key limitations. First, most methods rely on limited 2D cineloops, thereby ignoring widely available Spectral Doppler imaging that contains important complementary information about pressure gradients and blood flow abnormalities associated with AS. Second, obtaining labeled data is difficult. There are often far more unlabeled echocardiogram recordings available, but these remain underutilized by existing methods. To overcome these limitations, we introduce Semi-supervised Multimodal Multiple-Instance Learning (SMMIL), a new deep learning framework for automatic interpretation for structural heart diseases like AS. During training, SMMIL can combine a smaller labeled set and an abundant unlabeled set of both 2D and Doppler modalities to improve its classifier. When deployed, SMMIL can combine information from all available images to produce an accurate study-level diagnosis of this life-threatening condition. Experiments demonstrate that SMMIL outperforms recent alternatives, including two medical foundation models. 

Automated interpretation of ultrasound imaging of the heart (echocardiograms) could improve the detection and treatment of aortic stenosis (AS), a deadly heart disease. However, existing deep learning pipelines for assessing AS from echocardiograms have two key limitations. First, most methods rely on limited 2D cineloops, thereby ignoring widely available Spectral Doppler imaging that contains important complementary information about pressure gradients and blood flow abnormalities associated with AS. Second, obtaining labeled data is difficult. There are often far more unlabeled echocardiogram recordings available, but these remain underutilized by existing methods. To overcome these limitations, we introduce Semi-supervised Multimodal Multiple-Instance Learning (SMMIL), a new deep learning framework for automatic interpretation for structural heart diseases like AS. During training, SMMIL can combine a smaller labeled set and an abundant unlabeled set of both 2D and Doppler modalities to improve its classifier. When deployed, SMMIL can combine information from all available images to produce an accurate study-level diagnosis of this life-threatening condition. Experiments demonstrate that SMMIL outperforms recent alternatives, including two medical foundation models. 

To inform public health interventions, researchers have developed models to forecast opioid-related overdose mortality. These efforts often have limited overlap in the models and datasets employed, presenting challenges to assessing progress in this field. Furthermore, common error-based performance metrics, such as root mean squared error (RMSE), cannot directly assess a key modeling purpose: the identification of priority areas for interventions. We recommend a new intervention-aware performance metric, Percentage of Best Possible Reach (%BPR). We compare metrics for many published models across two distinct geographic settings, Cook County, Illinois and Massachusetts, assuming the budget to intervene in 100 census tracts out of 1000s in each setting. The top-performing models based on RMSE recommend areas that do not always reach the most possible overdose events. In Massachusetts, the top models preferred by %BPR could have reached 18 additional fatal overdoses per year in 2020-2021 compared to models favored by RMSE. In Cook County, the different metrics select similar top-performing models, yet other models with similar RMSE can have significant variation in %BPR. We further find that simple models often perform as well as recently published ones. We release open code and data for others to build upon.