Data-Type Assessment for Real-Time Hyperspectral Classification in Medical Imaging
Manuel Villa
(1)
,
Jaime Sancho
(1)
,
Guillermo Vazquez
(1)
,
Gonzalo Rosa
(1)
,
Gemma Urbanos
(1)
,
Alberto Martin-Perez
(1)
,
Pallab Sutradhar
(1)
,
Ruben Salvador
(2)
,
Miguel Chavarrias
(1)
,
Alfonso Lagares
(3)
,
Eduardo Juarez
(1)
,
Cesar Sanz
(1)
Ruben Salvador
- Fonction : Auteur
- PersonId : 185132
- IdHAL : ruben-salvador-perea
- ORCID : 0000-0002-0021-5808
Résumé
Real-time constraints in image processing applications often force their optimization using hardware accelerators. This is the case for intraoperative medical images used during surgical procedures. In this context, the challenge consists in processing large volumes of data while employing high complexity algorithms in a limited period of time. Newly developed algorithms must meet both quality-accurate and hardware-efficient characteristics. In this work, we have evaluated the impact of using different data types in a processing chain to classify tissues from hyperspectral video in surgical environments. The software was run on two different embedded CPU+GPU platforms. The results show an improvement in performance by up to 9 times without increasing power consumption by reducing the bit depth from 64 to 16. The impact these reduction have on quality has been measured analytically, by calculating the RMSE, and subjectively, by surveying neurosurgeons. In both cases the results show a minimal impact on the overall quality.
Domaines
Sciences de l'ingénieur [physics]Format du dépôt | Notice |
---|---|
Type de dépôt | Communication dans un congrès |
Titre |
en
Data-Type Assessment for Real-Time Hyperspectral Classification in Medical Imaging
|
Résumé |
en
Real-time constraints in image processing applications often force their optimization using hardware accelerators. This is the case for intraoperative medical images used during surgical procedures. In this context, the challenge consists in processing large volumes of data while employing high complexity algorithms in a limited period of time. Newly developed algorithms must meet both quality-accurate and hardware-efficient characteristics. In this work, we have evaluated the impact of using different data types in a processing chain to classify tissues from hyperspectral video in surgical environments. The software was run on two different embedded CPU+GPU platforms. The results show an improvement in performance by up to 9 times without increasing power consumption by reducing the bit depth from 64 to 16. The impact these reduction have on quality has been measured analytically, by calculating the RMSE, and subjectively, by surveying neurosurgeons. In both cases the results show a minimal impact on the overall quality.
|
Auteur(s) |
Manuel Villa
1
, Jaime Sancho
1
, Guillermo Vazquez
1
, Gonzalo Rosa
1
, Gemma Urbanos
1
, Alberto Martin-Perez
1
, Pallab Sutradhar
1
, Ruben Salvador
2
, Miguel Chavarrias
1
, Alfonso Lagares
3
, Eduardo Juarez
1
, Cesar Sanz
1
1
UPM -
Universidad Politécnica de Madrid
( 302798 )
- Calle Ramiro de Maeztu, 7, 28040 Madrid
- Espagne
2
IETR -
Institut d'Électronique et des Technologies du numéRique
( 1089025 )
- Campus de Beaulieu Bâtiment 11D 263 Av.Général Leclerc-CS 74205 35042 Rennes Cedex
- France
3
Hospital Universitario 12 de Octubre [Madrid]
( 314146 )
- Avda. de Córdoba, s/n 28041 Madrid
- Espagne
|
Langue du document |
Anglais
|
Source |
15th International Workshop on Design and Architectures for Signal and Image Processing (DASIP)
|
Vulgarisation |
Non
|
Comité de lecture |
Oui
|
Invité |
Non
|
Audience |
Internationale
|
Actes |
Oui
|
Date de publication |
2022
|
Titre de la collection |
DESIGN AND ARCHITECTURE FOR SIGNAL AND IMAGE PROCESSING, DASIP 2022
|
Titre du congrès |
15th International Workshop on Design and Architectures for Signal and Image Processing (DASIP)
|
Date début congrès |
2022-06-20
|
Date fin congrès |
2022-06-22
|
Ville |
Budapest
|
Pays |
Hongrie
|
Domaine(s) |
|
Éditeur commercial |
|
Éditeur scientifique |
|
Financement |
|
Mots-clés |
en
HSI, ML, tumor, video, embedded, GPU, real-time
|
DOI | 10.1007/978-3-031-12748-9_10 |
Loading...