Cisco Validated Design reference architecture for SAS Artificial Intelligence / Machine Learning

Principal Investigator

Name
Andrew Lockhart

Degrees
BA

Institution
SAS

Position Title
AI Partnership Strategist

Email
andrew.lockhart@sas.com

About this CDAS Project

Study
NLST (Learn more about this study)

Project ID
NLST-499

Initial CDAS Request Approval
Apr 13, 2019

Title
Cisco Validated Design reference architecture for SAS Artificial Intelligence / Machine Learning

Summary
The SAS Artificial Intelligence / Machine Learning team is working with SAS Partner, Cisco Systems, to build a Cisco Validated Design (https://www.cisco.com/c/en/us/solutions/design-zone.html) reference architecture for SAS AI/ML that is GPU-based and medical imaging-focused. In order to build this system effectively, we are seeking the largest available dataset of anonymized medical images for the SAS Computer Vision algorithms to process and train on. We identified the NLST data via the Cancer Imaging Archive as a prime candidate for this project because of its size. If approved, the SAS/Cisco team would load the NLST images into the reference system, train the SAS Computer Vision AI to gain familiarity with the images in order to be able to recognize/infer characteristics and then formally document all relevant performance metrics before dismantling the system and publishing the CVD. Any reference in the published CVD to the image data would be in the form of high-level performance metrics such as X number of images processed in Y amount of time, etc.

Aims

Primary: Build and document a reference architecture for SAS Computer Vision AI in a GPU-based Cisco environment
Secondary: Train the SAS Computer Vision AI on NLST images in order to test and fine tune the anomaly detection capabilities

Collaborators

Diana Shaw, SAS
Fijoy Vadakkumpadan, SAS
Brian Garrett, SAS
Scott Chastain, SAS
John Lambert, SAS
Vadi Bhatt, Cisco
Silesh Bijjahalli, Cisco