Features for detecting smoke in laparoscopic videos
Video-based smoke detection in laparoscopic surgery has different potential applications, such as the automatic addressing of surgical events associated with the electrocauterization task and the development of automatic smoke removal. In the literature, video-based smoke detection has been studied...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
De Gruyter
2017-09-01
|
Series: | Current Directions in Biomedical Engineering |
Subjects: | |
Online Access: | https://doi.org/10.1515/cdbme-2017-0110 |
_version_ | 1797848705939800064 |
---|---|
author | Jalal Nour Aldeen Alshirbaji Tamer Abdulbaki Mündermann Lars Möller Knut |
author_facet | Jalal Nour Aldeen Alshirbaji Tamer Abdulbaki Mündermann Lars Möller Knut |
author_sort | Jalal Nour Aldeen |
collection | DOAJ |
description | Video-based smoke detection in laparoscopic surgery has different potential applications, such as the automatic addressing of surgical events associated with the electrocauterization task and the development of automatic smoke removal. In the literature, video-based smoke detection has been studied widely for fire surveillance systems. Nevertheless, the proposed methods are insufficient for smoke detection in laparoscopic videos because they often depend on assumptions which rarely hold in laparoscopic surgery such as static camera. In this paper, ten visual features based on motion, texture and colour of smoke are proposed and evaluated for smoke detection in laparoscopic videos. These features are RGB channels, energy-based feature, texture features based on gray level co-occurrence matrix (GLCM), HSV colour space feature, features based on the detection of moving regions using optical flow and the smoke colour in HSV colour space. These features were tested on four laparoscopic cholecystectomy videos. Experimental observations show that each feature can provide valuable information in performing the smoke detection task. However, each feature has weaknesses to detect the presence of smoke in some cases. By combining all proposed features smoke with high and even low density can be identified robustly and the classification accuracy increases significantly. |
first_indexed | 2024-04-09T18:33:01Z |
format | Article |
id | doaj.art-65ab3b6e26e241068d0d7b8191edfc88 |
institution | Directory Open Access Journal |
issn | 2364-5504 |
language | English |
last_indexed | 2024-04-09T18:33:01Z |
publishDate | 2017-09-01 |
publisher | De Gruyter |
record_format | Article |
series | Current Directions in Biomedical Engineering |
spelling | doaj.art-65ab3b6e26e241068d0d7b8191edfc882023-04-11T17:07:14ZengDe GruyterCurrent Directions in Biomedical Engineering2364-55042017-09-013252152410.1515/cdbme-2017-0110cdbme-2017-0110Features for detecting smoke in laparoscopic videosJalal Nour Aldeen0Alshirbaji Tamer Abdulbaki1Mündermann Lars2Möller Knut3Furtwangen University, Institute of Technical Medicine, Germany Furtwangen University, Institute of Technical Medicine, Germany Karl-Storz GmbH, Germany Furtwangen University, Institute of Technical Medicine, Germany Video-based smoke detection in laparoscopic surgery has different potential applications, such as the automatic addressing of surgical events associated with the electrocauterization task and the development of automatic smoke removal. In the literature, video-based smoke detection has been studied widely for fire surveillance systems. Nevertheless, the proposed methods are insufficient for smoke detection in laparoscopic videos because they often depend on assumptions which rarely hold in laparoscopic surgery such as static camera. In this paper, ten visual features based on motion, texture and colour of smoke are proposed and evaluated for smoke detection in laparoscopic videos. These features are RGB channels, energy-based feature, texture features based on gray level co-occurrence matrix (GLCM), HSV colour space feature, features based on the detection of moving regions using optical flow and the smoke colour in HSV colour space. These features were tested on four laparoscopic cholecystectomy videos. Experimental observations show that each feature can provide valuable information in performing the smoke detection task. However, each feature has weaknesses to detect the presence of smoke in some cases. By combining all proposed features smoke with high and even low density can be identified robustly and the classification accuracy increases significantly.https://doi.org/10.1515/cdbme-2017-0110laparoscopic videossmoke detectionhsv colour space |
spellingShingle | Jalal Nour Aldeen Alshirbaji Tamer Abdulbaki Mündermann Lars Möller Knut Features for detecting smoke in laparoscopic videos Current Directions in Biomedical Engineering laparoscopic videos smoke detection hsv colour space |
title | Features for detecting smoke in laparoscopic videos |
title_full | Features for detecting smoke in laparoscopic videos |
title_fullStr | Features for detecting smoke in laparoscopic videos |
title_full_unstemmed | Features for detecting smoke in laparoscopic videos |
title_short | Features for detecting smoke in laparoscopic videos |
title_sort | features for detecting smoke in laparoscopic videos |
topic | laparoscopic videos smoke detection hsv colour space |
url | https://doi.org/10.1515/cdbme-2017-0110 |
work_keys_str_mv | AT jalalnouraldeen featuresfordetectingsmokeinlaparoscopicvideos AT alshirbajitamerabdulbaki featuresfordetectingsmokeinlaparoscopicvideos AT mundermannlars featuresfordetectingsmokeinlaparoscopicvideos AT mollerknut featuresfordetectingsmokeinlaparoscopicvideos |