A wireless narrowband imaging chip for capsule endoscope

Lan-Rong Dung*, Yin Yi Wu

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

26 Scopus citations


This paper presents a dual-mode capsule gastrointestinal endoscope device. An endoscope combined with a narrowband image (NBI), has been shown to be a superior diagnostic tool for early stage tissue neoplasms detection. Nevertheless, a wireless capsule endoscope with the narrowband imaging technology has not been presented in the market up to now. The narrowband image acquisition and power dissipation reduction are the main challenges of NBI capsule endoscope. In this paper, we present the first narrowband imaging capsule endoscope that can assist clinical doctors to effectively diagnose early gastrointestinal cancers, profited from our dedicated dual-mode complementary metaloxide semiconductor (CMOS) sensor. The dedicated dual-mode CMOS sensor can offer white-light and narrowband images. Implementation results show that the proposed 512 × 512 CMOS sensor consumes only 2 mA at a 3-V power supply. The average current of the NBI capsule with an 8-Mb/s RF transmitter is nearly 7 ∼ 8 mA that can continuously work for 6 ∼ 8 h with two 1.5-V 80-mAh button batteries while the frame rate is 2 fps. Experimental results on backside mucosa of a human tongue and pig's small intestine showed that the wireless NBI capsule endoscope can significantly improve the image quality, compared with a commercial-of-the-shelf capsule endoscope for gastrointestinal tract diagnosis.

Original languageEnglish
Article number5634124
Pages (from-to)462-468
Number of pages7
JournalIEEE Transactions on Biomedical Circuits and Systems
Issue number6 PART 2
StatePublished - 1 Dec 2010


  • Capsule endoscope
  • complementary metaloxide semiconductor (CMOS) sensor
  • gastrointestinal image
  • narrowband image (NBI)

Fingerprint Dive into the research topics of 'A wireless narrowband imaging chip for capsule endoscope'. Together they form a unique fingerprint.

Cite this