MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Effective Algorithm Sharing and Validating Using Serverless Cloud Computing In Radiation Oncology

Y Chen1 , J Xia2*, (1) University of Iowa, Iowa City, IA, (2) University Of Iowa Hospitals and Clinics, Iowa City, IA

Presentations

(Sunday, 7/29/2018) 3:00 PM - 6:00 PM

Room: Exhibit Hall

Purpose: To demonstrate the feasibility of validating and sharing computer algorithms base on the serverless cloud computing framework in order to accelerate scientific discovery, reduce software commercialization cycle, and improve algorithm robustness.

Methods: Serverless cloud computing allows to build and execute software applications without dealing with the time-consuming computer server management. In serverless computing, no computational resources are allocated or chargeable until a computational function is called. The serverless computing framework consists of API gateway, event handler, cloud computing resources, and cloud storage. As a proof of concept, we developed a radiation oncology nomenclature standardization algorithm, which is shared on the cloud by utilizing the serverless computing framework from Amazon Web Services (AWS). Users can access the algorithm using web based REST (representational state transfer) API. No software installation is required for the users to test and execute the algorithm. Performance was assessed by measuring the following criteria: 1) the algorithm execution time in the cloud vs. on the local workstation; 2) time delay between the “warm start� (executing an active function) and “cold start� (executing an inactive function). The algorithm was executed in both the AWS cloud and the local computer for 10 times.

Results: An average runtime of the standardization algorithm using the AWS cloud is 0.137 seconds with the cold start and 0.0851 seconds with the warm start. Comparing with the cloud based executing environment, the runtime using a local computer is 0.005 seconds, much faster than cloud based algorithm. The difference is largely due to the tome of transferring data cross the networks.

Conclusion: Serverless computing provides a practical solution to share and validate algorithms without exposing the source code. It can reduce the time from software development to software deployment. We demonstrate its feasibility by sharing a standardization algorithm using AWS serverless framework.

Keywords

Computer Software

Taxonomy

IM- Dataset analysis/biomathematics: Cloud Computing

Contact Email