Hyperspectral image segmentation: a preliminary study on the Oral and Dental Spectral Image Database (ODSI-DB)

Luis C. Garcia-Peraza Herrera; Conor Horgan; Sebastien Ourselin; Michael Ebner; Tom Vercauteren

doi:10.1080/21681163.2022.2160377

Hyperspectral image segmentation: a preliminary study on the Oral and Dental Spectral Image Database (ODSI-DB)

Luis C. Garcia-Peraza Herrera, Conor Horgan, Sebastien Ourselin, Michael Ebner, Tom Vercauteren

Research output: Contribution to journal › Article › peer-review

39 Downloads (Pure)

Abstract

Visual discrimination of clinical tissue types remains challenging, with traditional RGB imaging providing limited contrast for such tasks. Hyperspectral imaging (HSI) is a promising technology providing rich spectral information that can extend far beyond three-channel RGB imaging. Moreover, recently developed snapshot HSI cameras enable real-time imaging with significant potential for clinical applications. Despite this, the investigation into the relative performance of HSI over RGB imaging for semantic segmentation purposes has been limited, particularly in the context of medical imaging. Here we compare the performance of state-of-the-art deep learning image segmentation methods when trained on hyperspectral images, RGB images, hyperspectral pixels (minus spatial context), and RGB pixels (disregarding spatial context). To achieve this, we employ the recently released Oral and Dental Spectral Image Database (ODSI-DB), which consists of 215 manually segmented dental reflectance spectral images with 35 different classes across 30 human subjects. The recent development of snapshot HSI cameras has made real-time clinical HSI a distinct possibility, though successful application requires a comprehensive understanding of the additional information HSI offers. Our work highlights the relative importance of spectral resolution, spectral range, and spatial information to both guide the development of HSI cameras and inform future clinical HSI applications.

Original language	English
Pages (from-to)	1290-1298
Number of pages	9
Journal	Frontiers in Robotics and AI
Volume	11
Issue number	4
Early online date	23 Jan 2023
DOIs	https://doi.org/10.1080/21681163.2022.2160377
Publication status	Published - 2023

Access to Document

10.1080/21681163.2022.2160377Licence: CC BY

Hyperspectral image segmentation a preliminary study on the Oral and Dental Spectral Image Database ODSI DBFinal published version, 4.29 MBLicence: CC BY

Cite this

@article{f8d1e56652134b8db4e48664e89f0590,

title = "Hyperspectral image segmentation: a preliminary study on the Oral and Dental Spectral Image Database (ODSI-DB)",

abstract = "Visual discrimination of clinical tissue types remains challenging, with traditional RGB imaging providing limited contrast for such tasks. Hyperspectral imaging (HSI) is a promising technology providing rich spectral information that can extend far beyond three-channel RGB imaging. Moreover, recently developed snapshot HSI cameras enable real-time imaging with significant potential for clinical applications. Despite this, the investigation into the relative performance of HSI over RGB imaging for semantic segmentation purposes has been limited, particularly in the context of medical imaging. Here we compare the performance of state-of-the-art deep learning image segmentation methods when trained on hyperspectral images, RGB images, hyperspectral pixels (minus spatial context), and RGB pixels (disregarding spatial context). To achieve this, we employ the recently released Oral and Dental Spectral Image Database (ODSI-DB), which consists of 215 manually segmented dental reflectance spectral images with 35 different classes across 30 human subjects. The recent development of snapshot HSI cameras has made real-time clinical HSI a distinct possibility, though successful application requires a comprehensive understanding of the additional information HSI offers. Our work highlights the relative importance of spectral resolution, spectral range, and spatial information to both guide the development of HSI cameras and inform future clinical HSI applications.",

author = "{Garcia-Peraza Herrera}, {Luis C.} and Conor Horgan and Sebastien Ourselin and Michael Ebner and Tom Vercauteren",

note = "Funding Information: This work was supported by core and project Publisher Copyright: {\textcopyright} 2023 Crown Copyright. Funding Information: The work was supported by the Horizon 2020 Framework Programme [101016985]; Innovate UK [75124]; EPSRC [NS/A000027/1]; Royal Academy of Engineering [RCSRF1819\7\34]; Wellcome Trust [WT101957,WT203148/Z/16/Z]. Publisher Copyright: {\textcopyright} 2023 Crown Copyright. Reproduced with the permission of the Controller of His Majesty{\textquoteright}s Stationery Office and Department of Surgical & Interventional Engineering, School of Biomedical Engineering & Imaging Sciences. Published by Informa UK Limited, trading as Taylor & Francis Group.",

year = "2023",

doi = "10.1080/21681163.2022.2160377",

language = "English",

volume = "11",

pages = "1290--1298",

journal = "Front. Robot. AI",

issn = "2296-9144",

publisher = "Frontiers Media S.A.",

number = "4",

}

TY - JOUR

T1 - Hyperspectral image segmentation

T2 - a preliminary study on the Oral and Dental Spectral Image Database (ODSI-DB)

AU - Garcia-Peraza Herrera, Luis C.

AU - Horgan, Conor

AU - Ourselin, Sebastien

AU - Ebner, Michael

AU - Vercauteren, Tom

N1 - Funding Information: This work was supported by core and project Publisher Copyright: © 2023 Crown Copyright. Funding Information: The work was supported by the Horizon 2020 Framework Programme [101016985]; Innovate UK [75124]; EPSRC [NS/A000027/1]; Royal Academy of Engineering [RCSRF1819\7\34]; Wellcome Trust [WT101957,WT203148/Z/16/Z]. Publisher Copyright: © 2023 Crown Copyright. Reproduced with the permission of the Controller of His Majesty’s Stationery Office and Department of Surgical & Interventional Engineering, School of Biomedical Engineering & Imaging Sciences. Published by Informa UK Limited, trading as Taylor & Francis Group.

PY - 2023

Y1 - 2023

N2 - Visual discrimination of clinical tissue types remains challenging, with traditional RGB imaging providing limited contrast for such tasks. Hyperspectral imaging (HSI) is a promising technology providing rich spectral information that can extend far beyond three-channel RGB imaging. Moreover, recently developed snapshot HSI cameras enable real-time imaging with significant potential for clinical applications. Despite this, the investigation into the relative performance of HSI over RGB imaging for semantic segmentation purposes has been limited, particularly in the context of medical imaging. Here we compare the performance of state-of-the-art deep learning image segmentation methods when trained on hyperspectral images, RGB images, hyperspectral pixels (minus spatial context), and RGB pixels (disregarding spatial context). To achieve this, we employ the recently released Oral and Dental Spectral Image Database (ODSI-DB), which consists of 215 manually segmented dental reflectance spectral images with 35 different classes across 30 human subjects. The recent development of snapshot HSI cameras has made real-time clinical HSI a distinct possibility, though successful application requires a comprehensive understanding of the additional information HSI offers. Our work highlights the relative importance of spectral resolution, spectral range, and spatial information to both guide the development of HSI cameras and inform future clinical HSI applications.

AB - Visual discrimination of clinical tissue types remains challenging, with traditional RGB imaging providing limited contrast for such tasks. Hyperspectral imaging (HSI) is a promising technology providing rich spectral information that can extend far beyond three-channel RGB imaging. Moreover, recently developed snapshot HSI cameras enable real-time imaging with significant potential for clinical applications. Despite this, the investigation into the relative performance of HSI over RGB imaging for semantic segmentation purposes has been limited, particularly in the context of medical imaging. Here we compare the performance of state-of-the-art deep learning image segmentation methods when trained on hyperspectral images, RGB images, hyperspectral pixels (minus spatial context), and RGB pixels (disregarding spatial context). To achieve this, we employ the recently released Oral and Dental Spectral Image Database (ODSI-DB), which consists of 215 manually segmented dental reflectance spectral images with 35 different classes across 30 human subjects. The recent development of snapshot HSI cameras has made real-time clinical HSI a distinct possibility, though successful application requires a comprehensive understanding of the additional information HSI offers. Our work highlights the relative importance of spectral resolution, spectral range, and spatial information to both guide the development of HSI cameras and inform future clinical HSI applications.

UR - http://www.scopus.com/inward/record.url?scp=85147041503&partnerID=8YFLogxK

U2 - 10.1080/21681163.2022.2160377

DO - 10.1080/21681163.2022.2160377

M3 - Article

SN - 2296-9144

VL - 11

SP - 1290

EP - 1298

JO - Front. Robot. AI

JF - Front. Robot. AI

IS - 4

ER -

Hyperspectral image segmentation: a preliminary study on the Oral and Dental Spectral Image Database (ODSI-DB)

Abstract

Access to Document

Other files and links

Fingerprint

Cite this