Search

Measurements of liver volume from MR images can be valuable for both clinical and research applications. Automated methods using convolutional neural networks have been used successfully for this using a variety of different MR image types as input. In this work, we sought to determine which types of magnetic resonance images give the best performance when used to train convolutional neural networks for liver segmentation and volumetry. Abdominal MRI scans were performed at 3 Tesla on 42 adolescents with obesity. Scans included Dixon imaging (giving water, fat, and T2* images) and low-resolution T2-weighted scout images. Multiple convolutional neural network models using a 3D U-Net architecture were trained with different input images. Whole-liver manual segmentations were used for reference. Segmentation performance was measured using the Dice similarity coefficient (DSC) and 95% Hausdorff distance. Liver volume accuracy was evaluated using bias, precision, intraclass correlation coefficient, normalized root mean square error (NRMSE), and Bland-Altman analyses. The models trained using both water and fat images performed best, giving DSC = 0.94 and NRMSE = 4.2%. Models trained without the water image as input all performed worse, including in participants with elevated liver fat. Models using the T2-weighted scout images underperformed the Dixon-based models, but provided acceptable performance (DSC ≥ 0.92, NMRSE ≤ 6.6%) for use in longitudinal pediatric obesity interventions. The model using Dixon water and fat images as input gave the best performance, with results comparable to inter-reader variability and state-of-the-art methods.

Application of data science tools, techniques, and principles have increased within the field of library and information science. This trend is especially noticeable in academic libraries where strategic priorities around data science initiatives have been created to further support and add value to the research enterprises at their institutions. This chapter seeks to highlight a global outlook on how data science has been addressed in library and information studies via case studies in areas including digital humanities, machine learning, and visual analytics. Increasing awareness of how partners across the globe are addressing data science needs at their institutions can help raise visibility of how data science can be infused and utilized within a variety of contexts.

OBJECTIVE: We sought to evaluate the effect of 52 weeks of exenatide extended release (XR) on maintenance of meal replacement therapy induced BMI reduction in adolescents with severe obesity. METHODS: In this randomized, double-blind, placebo-controlled trial, 100 participants aged 12

Measurements of liver volume from MR images can be valuable for both clinical and research applications. Automated methods using convolutional neural networks have been used successfully for this using a variety of different MR image types as input. In this work, we sought to determine which types of magnetic resonance images give the best performance when used to train convolutional neural networks for liver segmentation and volumetry.Abdominal MRI scans were performed at 3 Tesla on 42 adolescents with obesity. Scans included Dixon imaging (giving water, fat, and T2* images) and low-resolution T2-weighted anatomical scans. Multiple convolutional neural network models using a 3D U-Net architecture were trained with different input images. Whole-liver manual segmentations were used for reference. Segmentation performance was measured using the Dice similarity coefficient (DSC) and 95% Hausdorff distance. Liver volume accuracy was evaluated using bias, precision, and normalized root mean square error (NRMSE).The models trained using both water and fat images performed best, giving DSC = 0.94 and NRMSE = 4.2%. Models trained without the water image as input all performed worse, including in participants with elevated liver fat. Models using the T2-weighted anatomical images underperformed the Dixon-based models, but provided acceptable performance (DSC ≥ 0.92, NMRSE ≤ 6.6%) for use in longitudinal pediatric obesity interventions. The model using Dixon water and fat images as input gave the best performance, with results comparable to inter-reader variability and state-of-the-art methods.

AB The activPAL is an accelerometer-based monitor worn on the thigh that classifies daily activities into three categories (sitting/lying down, standing, and stepping). The monitor discriminates between sitting/lying and the upright position by detecting the inclination of the thigh. It detects stepping from the acceleration versus time wave form. However, a current limitation of the activPAL is that it does not discriminate between sitting and lying down. Purpose: This study aimed to determine whether placing a second activPAL monitor on the torso would allow the detection of seated versus lying postures. Methods: Fifteen healthy adults (18-55 yr of age) wore an activPAL on the right thigh and another activPAL over the right rib cage. Both monitors were synchronized and initialized to record data in 15-s epochs. Participants performed a semistructured routine of activities for 3 min each. Activities included lying down (while supine, prone, and on the side), sitting, standing, sweeping, treadmill walking at 3 mph, and treadmill running at 6 mph. The spatial orientation of the thigh and chest monitors was used to determine body posture, and the activPAL on the thigh was used to detect ambulation. Results: The use of two activPAL devices enabled four behaviors to be accurately classified. The percentages of observations that were classified accurately were as follows: lying down (100%), sitting (100%), standing/light activity in the upright position (90.8%), and stepping (100%). Conclusions: The current method allows researchers to obtain more detailed information on postural allocation compared with that in the use of a single activPAL on the thigh