Active multitask learning with uncertainty-weighted loss for coronary calcium scoring.

PURPOSE: The coronary artery calcification (CAC) score is an independent marker for the risk of cardiovascular events. Automatic methods for quantifying CAC could reduce workload and assist radiologists in clinical decision-making. However, large annotated datasets are needed for training to achieve very good model performance, which is an expensive process and requires expert knowledge. The number of training data required can be reduced in an active learning scenario, which requires only the most informative samples to be labeled. Multitask learning techniques can improve model performance by joint learning of multiple related tasks and extraction of shared informative features. METHODS: We propose an uncertainty-weighted multitask learning model for coronary calcium scoring in electrocardiogram-gated (ECG-gated), noncontrast-enhanced cardiac calcium scoring CT. The model was trained to solve the two tasks of coronary artery region segmentation (weak labels) and coronary artery calcification segmentation (strong labels) simultaneously in an active learning scenario to improve model performance and reduce the number of samples needed for training. We compared our model with a single-task U-Net and a sequential-task model as well as other state-of-the-art methods. The model was evaluated on 1275 individual patients in three different datasets (DISCHARGE, CADMAN, orCaScore), and the relationship between model performance and various influencing factors (image noise, metal artifacts, motion artifacts, image quality) was analyzed. RESULTS: Joint learning of multiclass coronary artery region segmentation and binary coronary calcium segmentation improved calcium scoring performance. Since shared information can be learned from both tasks for complementary purposes, the model reached optimal performance with only 12% of the training data and one-third of the labeling time in an active learning scenario. We identified image noise as one of the most important factors influencing model performance along with anatomical abnormalities and metal artifacts. CONCLUSIONS: Our multitask learning approach with uncertainty-weighted loss improves calcium scoring performance by joint learning of shared features and reduces labeling costs when trained in an active learning scenario.

Audio-guided self-hypnosis for reduction of claustrophobia during MR imaging: results of an observational 2-group study.

OBJECTIVES: To evaluate the influence of audio-guided self-hypnosis on claustrophobia in a high-risk cohort undergoing magnetic resonance (MR) imaging. METHODS: In this prospective observational 2-group study, 55 patients (69% female, mean age 53.6 ± 13.9) used self-hypnosis directly before imaging. Claustrophobia included premature termination, sedation, and coping actions. The claustrophobia questionnaire (CLQ) was completed before self-hypnosis and after MR imaging. Results were compared to a control cohort of 89 patients examined on the same open MR scanner using logistic regression for multivariate analysis. Furthermore, patients were asked about their preferences for future imaging. RESULTS: There was significantly fewer claustrophobia in the self-hypnosis group (16%; 9/55), compared with the control group (43%; 38/89; odds ratio .14; p = .001). Self-hypnosis patients also needed less sedation (2% vs 16%; 1/55 vs 14/89; odds ratio .1; p = .008) and non-sedation coping actions (13% vs 28%; 7/55 vs 25/89; odds ratio .3; p = .02). Self-hypnosis did not influence the CLQ results measured before and after MR imaging (p = .79). Self-hypnosis reduced the frequency of claustrophobia in the subgroup of patients above an established CLQ cut-off of .33 from 47% (37/78) to 18% (9/49; p = .002). In the subgroup below the CLQ cut-off of 0.33, there were no significant differences (0% vs 9%, 0/6 vs 1/11; p = 1.0). Most patients (67%; 35/52) preferred self-hypnosis for future MR examinations. CONCLUSIONS: Self-hypnosis reduced claustrophobia in high-risk patients undergoing imaging in an open MR scanner and might reduce the need for sedation and non-sedation coping actions. KEY POINTS: • Forty percent of the patients at high risk for claustrophobia may also experience a claustrophobic event in an open MR scanner. • Self-hypnosis while listening to an audio in the waiting room before the examination may reduce claustrophobic events in over 50% of patients with high risk for claustrophobia. • Self-hypnosis may also reduce the need for sedation and other time-consuming non-sedation coping actions and is preferred by high-risk patients for future examinations.