Cover Image

Clinically guided trainable soft attention for early detection of oral cancer

Oral cancer disproportionately affects low- and middle-income countries, where a lack of access to appropriate medical care contributes towards late disease presentation. Using artificial intelligence to facilitate the automated identification of high-risk oral lesions can improve patient survival r...

Full description

Saved in:
Bibliographic Details
Main Authors: Welikala, Roshan Alex, Remagnino, Paolo, Lim, Jian Han, Chan, Chee Seng, Rajendran, Senthilmani, Kallarakkal, Thomas George, Mohd Zain, Rosnah, Jayasinghe, Ruwan Duminda, Rimal, Jyotsna, Kerr, Alexander Ross, Amtha, Rahmi, Patil, Karthikeya, Tilakaratne, Wanninayake Mudiyanselage, Cheong, Sok Ching, Barman, Sarah Ann
Format: Conference or Workshop Item
Published: 2021
Subjects:
RK Dentistry
Online Access:http://eprints.um.edu.my/35582/
Tags: Add Tag
No Tags, Be the first to tag this record!
id my.um.eprints.35582
record_format eprints
spelling my.um.eprints.355822023-10-24T06:23:50Z http://eprints.um.edu.my/35582/ Clinically guided trainable soft attention for early detection of oral cancer Welikala, Roshan Alex Remagnino, Paolo Lim, Jian Han Chan, Chee Seng Rajendran, Senthilmani Kallarakkal, Thomas George Mohd Zain, Rosnah Jayasinghe, Ruwan Duminda Rimal, Jyotsna Kerr, Alexander Ross Amtha, Rahmi Patil, Karthikeya Tilakaratne, Wanninayake Mudiyanselage Cheong, Sok Ching Barman, Sarah Ann RK Dentistry Oral cancer disproportionately affects low- and middle-income countries, where a lack of access to appropriate medical care contributes towards late disease presentation. Using artificial intelligence to facilitate the automated identification of high-risk oral lesions can improve patient survival rates. With image classification using oral cavity images and other forms of medical images, the information to be classified can often be extremely localized. To address this problem, we propose the use of convolutional neural networks with trainable soft attention. Further to this, we incorporate the use of localization loss to penalize the difference between attention maps and clinically annotated mask. This effectively allows clinicians to help guide soft attention. Improvements to the baseline were made, with an accuracy of 0.8333 and a ROC AUC of 0.8632, which equates to increases of 0.0245 and 0.0394, respectively. This accuracy corresponds to a sensitivity of 0.8469 and a specificity of 0.8208. Perhaps of more importance, is a model that demonstrates better capability at paying attention to the lesions in its decision making. Furthermore, visualizing resulting attention maps can help to strengthen clinical confidence in AI decision making. © 2021, Springer Nature Switzerland AG. 2021 Conference or Workshop Item PeerReviewed Welikala, Roshan Alex and Remagnino, Paolo and Lim, Jian Han and Chan, Chee Seng and Rajendran, Senthilmani and Kallarakkal, Thomas George and Mohd Zain, Rosnah and Jayasinghe, Ruwan Duminda and Rimal, Jyotsna and Kerr, Alexander Ross and Amtha, Rahmi and Patil, Karthikeya and Tilakaratne, Wanninayake Mudiyanselage and Cheong, Sok Ching and Barman, Sarah Ann (2021) Clinically guided trainable soft attention for early detection of oral cancer. In: 19th International Conference on Computer Analysis of Images and Patterns, CAIP 2021, 28 - 30 September 2021, Virtual, Online.
institution Universiti Malaya
building UM Library
collection Institutional Repository
continent Asia
country Malaysia
content_provider Universiti Malaya
content_source UM Research Repository
url_provider http://eprints.um.edu.my/
topic RK Dentistry
spellingShingle RK Dentistry
Welikala, Roshan Alex
Remagnino, Paolo
Lim, Jian Han
Chan, Chee Seng
Rajendran, Senthilmani
Kallarakkal, Thomas George
Mohd Zain, Rosnah
Jayasinghe, Ruwan Duminda
Rimal, Jyotsna
Kerr, Alexander Ross
Amtha, Rahmi
Patil, Karthikeya
Tilakaratne, Wanninayake Mudiyanselage
Cheong, Sok Ching
Barman, Sarah Ann
Clinically guided trainable soft attention for early detection of oral cancer
description Oral cancer disproportionately affects low- and middle-income countries, where a lack of access to appropriate medical care contributes towards late disease presentation. Using artificial intelligence to facilitate the automated identification of high-risk oral lesions can improve patient survival rates. With image classification using oral cavity images and other forms of medical images, the information to be classified can often be extremely localized. To address this problem, we propose the use of convolutional neural networks with trainable soft attention. Further to this, we incorporate the use of localization loss to penalize the difference between attention maps and clinically annotated mask. This effectively allows clinicians to help guide soft attention. Improvements to the baseline were made, with an accuracy of 0.8333 and a ROC AUC of 0.8632, which equates to increases of 0.0245 and 0.0394, respectively. This accuracy corresponds to a sensitivity of 0.8469 and a specificity of 0.8208. Perhaps of more importance, is a model that demonstrates better capability at paying attention to the lesions in its decision making. Furthermore, visualizing resulting attention maps can help to strengthen clinical confidence in AI decision making. © 2021, Springer Nature Switzerland AG.
format Conference or Workshop Item
author Welikala, Roshan Alex
Remagnino, Paolo
Lim, Jian Han
Chan, Chee Seng
Rajendran, Senthilmani
Kallarakkal, Thomas George
Mohd Zain, Rosnah
Jayasinghe, Ruwan Duminda
Rimal, Jyotsna
Kerr, Alexander Ross
Amtha, Rahmi
Patil, Karthikeya
Tilakaratne, Wanninayake Mudiyanselage
Cheong, Sok Ching
Barman, Sarah Ann
author_facet Welikala, Roshan Alex
Remagnino, Paolo
Lim, Jian Han
Chan, Chee Seng
Rajendran, Senthilmani
Kallarakkal, Thomas George
Mohd Zain, Rosnah
Jayasinghe, Ruwan Duminda
Rimal, Jyotsna
Kerr, Alexander Ross
Amtha, Rahmi
Patil, Karthikeya
Tilakaratne, Wanninayake Mudiyanselage
Cheong, Sok Ching
Barman, Sarah Ann
author_sort Welikala, Roshan Alex
title Clinically guided trainable soft attention for early detection of oral cancer
title_short Clinically guided trainable soft attention for early detection of oral cancer
title_full Clinically guided trainable soft attention for early detection of oral cancer
title_fullStr Clinically guided trainable soft attention for early detection of oral cancer
title_full_unstemmed Clinically guided trainable soft attention for early detection of oral cancer
title_sort clinically guided trainable soft attention for early detection of oral cancer
publishDate 2021
url http://eprints.um.edu.my/35582/
_version_ 1781704485504024576
score 13.211869

Similar Items