Using AI to Help Save Lives: A Data Driven Approach for Intracranial Hemorrhage Detection

ID 标签 689420
已更新 4/2/2020
版本 Latest
公共

author-image

作者

Authors:

Hector Robles, Cord Meados, Miguel Daniel Reyes Martinez, Gabriel Briones Sayeg,
David Esparza Borquez, Hugo Soto, Eduardo I Rascon Garcia, Tiejun Li, Jose A Lamego,
Geronimo Orozco, Luis F Ponce Navarro, Ali R Khan, Beth Dean, Rahul Unnikrishnan Nair

Contributors:

Esther John, Nectar Kirkiris

(All authors and contributors are employees of Intel Corporation.)

Abstract

Artificial intelligence (AI) is a key driver of the Fourth Industrial Revolution, and it promises to change the way we work across many industries. In health care, AI can provide insights into patient data, whether analyzing records or examining images. Despite all of the advancements in artificial intelligence, developers have been forced to do the integration across solutions and frameworks by themselves – a difficult task that distracts from the development effort. In this paper, we focus on solving two problems facing the domains of medical diagnosis and artificial intelligence.  First, we designed an AI training pipeline to detect intracranial hemorrhage (ICH), a serious condition often caused by traumatic brain injuries. ICH must be diagnosed and treated as quickly as possible to avoid disability or death of the patient. Second, we tackled the complexity of creating an AI pipeline with multiple software frameworks, configurations, and dependencies. Our solution was to use the System Stacks for Linux* OS, a purpose-built collection of containers that provide integrated, and tuned AI frameworks. We created an end-to-end solution that can take computerized tomography (CT) images of the brain, process them to extract relevant data, and provide feedback to the radiologist or lab technician. The feedback indicates if ICH is present, helping them to focus their examination on the relevant areas of the scan. Our solution can be deployed as a secure cloud service or can work on premises, according to need.  This will help enable remote clinics to take advantage of the system and provide efficiencies of scale.

Using the System Stacks for Linux OS for rapid development and deployment of real-world use cases reduces developer and infrastructure complexity. Developers are able to devote their time to the problem they are solving, rather than spending that time installing, configuring and integrating the AI frameworks and tools we have used here. With this use case, we can see how the stacks can be combined into a cohesive pipeline to manage the end-to-end solution and assist radiologists and lab technicians with detecting intracranial hemorrhage (ICH). The solution described here can be used directly or as a template to create pipelines to assist with different image recognition problems.  The System Stacks for Linux OS are optimized and tuned for performance out of the box for Intel® Xeon® scalable platforms, which simplifies the task of integrating the frameworks needed for a complicated pipeline like this one.

"