Accelerating FPGA-Implementations for Mobile Medical Devices with high-level AI libraries: an Object Detection Model for Colorectal Polyp Images

Sunyoung Yoo, Yunlong Wang, Jan-Matthias Braun, Esmaeil Nadimi

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

Abstract

Colorectal polyps are an important cause of colon cancer and can be effectively detected by colonoscopy or Wireless Capsule Endoscopy (WCE). Up to now, many studies have applied Deep Learning (DL) to polyp detection and have obtained promising results, but all the algorithms are run on GPUs. They cannot be transferred to the fast-growing segment of mobile medical devices which require low power uptake. To bridge this gap, we designed a polyp detection algorithm for a Field Programmable Gate Array (FPGA). We present the detailed procedure of Deep Neural Network (DNN) design and the FPGA implementation and show that the FPGA-based polyp detection outperforms the GPU-based one in terms of energy efficiency and lower latency. We thus demonstrate the feasibility of an application scenario for energy-efficient mobile medical devices without highly specialized FPGA skills.

Original languageEnglish
Title of host publication2021 IEEE International Conference on Imaging Systems and Techniques (IST)
PublisherIEEE
Publication date27. Dec 2021
Pages1-6
ISBN (Electronic)9781728173719
DOIs
Publication statusPublished - 27. Dec 2021
Event2021 IEEE International Conference on Imaging Systems and Techniques Virtual -
Duration: 24. Aug 20212. Sept 2021

Conference

Conference2021 IEEE International Conference on Imaging Systems and Techniques Virtual
Period24/08/202102/09/2021

Keywords

  • Deep learning
  • FPGA
  • Mobile medical devices
  • Polyp detection

Fingerprint

Dive into the research topics of 'Accelerating FPGA-Implementations for Mobile Medical Devices with high-level AI libraries: an Object Detection Model for Colorectal Polyp Images'. Together they form a unique fingerprint.

Cite this