Raw Data and Results from Automatic Murine Cardiac Ultrasound and Photoacoustic Image Segmentations

<p>The purpose of this project was to apply deep learning via implementation of U-Net to automatically and more efficiently segment the anterior myocardium of murine cardiovascular ultrasound and photoacoustic data in order to calculate radial strain and oxygen saturation. By doing this, segmentation time improved 300-fold and the deep neural network used was able to perform segmentation on both healthy and infarcted myocardium. The results of this project show that deep learning can be used to extract physiological parameters from ultrasound and photoacoustic data in less time to monitor cardiac health. The content in this publication includes raw data and results from running the deep neural network on the validation set of the ultrasound and photoacoustic images.</p>