Category Archives: NVIDIA

Segmentation and LV localization Based Approaches

By | Booz Allen, NVIDIA | No Comments
In our last blog post we described an end-to-end deep learning solution to this challenge. By “end-to-end” we mean that the raw pixels constituting a SAX study for an individual patient were fed into a convolutional neural network (ConvNet) and predicted left ventricle (LV) systolic and diastolic CDFs came out the other end – the only other processing that took place was the zero mean unit variance (ZMUV) pre-processing of the images. Whilst this approach to the problem is elegant in its simplicity, it is also a very challenging function for a neural network to learn. This is because there is no explicit training signal for the area of the left ventricle that should be measured from each image, just the whole volume for the SAX study. Read More

Building and Working on a Dispersed Team

By | Booz Allen, NVIDIA | No Comments

This year is the first time that Booz Allen and NVIDIA have partnered to enter a team into the Data Science Bowl. Our goal for this combined team was to share some of our successes and challenges along the way, as well as to provide insight into how to approach this type of competition. We’ve been able to post updates about our progress, respond to questions on the Kaggle forums, and help other teams find new ways of looking at the problem. Of course, we’re also hoping that by combining our talent and resources we will be able to come up with a top solution – even if we’re not eligible for the prize money. Read More

Image Preprocessing: The Challenges and Approach

By | Booz Allen, NVIDIA | No Comments
The dataset for the 2016 Data Science Bowl presents several challenges for automated exploitation. As the images were collected in a real world setting, with several types of sensors, there is a great deal of variation from patient to patient with respect to image orientation, pixel spacing, and intensity scaling. All of these factors should be dealt as part of any competitive solution; while they may not be required for a good solution, a winning design requires every last bit of information to be squeezed out of the data. Read More