Presentation 2014-11-26
Visual-Word Feature Transformation Architecture for Computer-Aided Diagnosis using Colorectal Endoscopic Images with NBI Magnification
Kouki SUGI, Tetsushi KOIDE, Anh-Tuan HOANG, Takumi OKAMOTO, Tatsuya SHIMIZU, Toru TAMAKI, Bisser Raytchev, Kazufumi KANEDA, Yoko KOMINAMI, Shigeto YOSHIDA, Shinji TANAKA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) With the increase of colorectal cancer patients in recent years, the computer-aided diagnosis (CAD) system which supports doctor's diagnosis effectively in the diagnosis of colon cancer is required. Our research describe the support system for colorectal endoscopic images with narrow band imaging (NBI) magnification which identify the pathology type from local features in NBI endoscopic image. In order to meet the real time processing in which a full high definition (Full HD) endoscopic image must be identified within one second, the feature transformation which change feature quantity to VW histogram is implemented on hardware. For implementation on FPGA which has the resource limitations, we propose the architecture that uses Manhattan distance in class searching, and enables high speed processing by dynamic computation of normalization. As a result, it allows the feature transformation to complete its work in about 334 msec on the full HD image. More over, the architecture can improve the processing time to about 43 msec by pipelining.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Colorectal Endoscopic Images / Full HD / Computer-Aided Diagnosis (CAD) / NBI (Narrow Band Imaging) / Feature Transformation / Visual-Word (VW) / Manhattan distance / FPGA Hardware Implementation
Paper # VLD2014-77,DC2014-31
Date of Issue

Conference Information
Committee DC
Conference Date 2014/11/19(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Dependable Computing (DC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Visual-Word Feature Transformation Architecture for Computer-Aided Diagnosis using Colorectal Endoscopic Images with NBI Magnification
Sub Title (in English)
Keyword(1) Colorectal Endoscopic Images
Keyword(2) Full HD
Keyword(3) Computer-Aided Diagnosis (CAD)
Keyword(4) NBI (Narrow Band Imaging)
Keyword(5) Feature Transformation
Keyword(6) Visual-Word (VW)
Keyword(7) Manhattan distance
Keyword(8) FPGA Hardware Implementation
1st Author's Name Kouki SUGI
1st Author's Affiliation Research Institute for Nanodevice and Bio System (RNBS), Hiroshima University()
2nd Author's Name Tetsushi KOIDE
2nd Author's Affiliation Research Institute for Nanodevice and Bio System (RNBS), Hiroshima University
3rd Author's Name Anh-Tuan HOANG
3rd Author's Affiliation Research Institute for Nanodevice and Bio System (RNBS), Hiroshima University
4th Author's Name Takumi OKAMOTO
4th Author's Affiliation Research Institute for Nanodevice and Bio System (RNBS), Hiroshima University
5th Author's Name Tatsuya SHIMIZU
5th Author's Affiliation Research Institute for Nanodevice and Bio System (RNBS), Hiroshima University
6th Author's Name Toru TAMAKI
6th Author's Affiliation Research Institute for Nanodevice and Bio System (RNBS), Hiroshima University
7th Author's Name Bisser Raytchev
7th Author's Affiliation Graduate School of Engineering
8th Author's Name Kazufumi KANEDA
8th Author's Affiliation Research Institute for Nanodevice and Bio System (RNBS), Hiroshima University:Graduate School of Engineering
9th Author's Name Yoko KOMINAMI
9th Author's Affiliation Graduate School of Biomedical Sciences, Hiroshima University
10th Author's Name Shigeto YOSHIDA
10th Author's Affiliation Research Institute for Nanodevice and Bio System (RNBS), Hiroshima University:Graduate School of Biomedical Sciences, Hiroshima University
11th Author's Name Shinji TANAKA
11th Author's Affiliation Graduate School of Biomedical Sciences, Hiroshima University
Date 2014-11-26
Paper # VLD2014-77,DC2014-31
Volume (vol) vol.114
Number (no) 329
Page pp.pp.-
#Pages 6
Date of Issue