Attention-guided convolutional network for bias-mitigated and interpretable oral lesion classification
Open Access
- 30 December 2024
- journal article
- research article
- Published by Springer Nature in Scientific Reports
- Vol. 14 (1) , 1-12
Abstract
Accurate diagnosis of oral lesions, early indicators of oral cancer, is a complex clinical challenge. Recent advances in deep learning have demonstrated potential in supporting clinical decisions. This paper introduces a deep learning model for classifying oral lesions, focusing on accuracy, interpretability, and reducing dataset bias. The model integrates three components: (i) a Classification Stream, utilizing a CNN to categorize images into 16 lesion types (baseline model), (ii) a Guidance Stream, which aligns class activation maps with clinically relevant areas using ground truth segmentation masks (GAIN model), and (iii) an Anatomical Site Prediction Stream, improving interpretability by predicting lesion location (GAIN+ASP model). The development dataset comprised 2765 intra-oral digital images of 16 lesion types from 1079 patients seen at an oral pathology clinic between 1999 and 2021. The GAIN model demonstrated a 7.2% relative improvement in accuracy over the baseline for 16-class classification, with superior class-specific balanced accuracy and AUC scores. Additionally, the GAIN model enhanced lesion localization and improved the alignment between attention maps and ground truth. The proposed models also exhibited greater robustness against dataset bias, as shown in ablation studies.Keywords
Funding Information
- Canadian Institutes of Health Research (PJT-438778)
This publication has 52 references indexed in Scilit:
- Artificial intelligence for classification and detection of oral mucosa lesions on photographs: a systematic review and meta-analysisClinical Oral Investigations, 2024
- Clinically Guided Trainable Soft Attention for Early Detection of Oral Cancer2021
- Attention Guided Anomaly Localization in Images2020
- Deep Learning in Medicine—Promise, Progress, and ChallengesJAMA Internal Medicine, 2019
- The importance of interpretability and visualization in machine learning for applications in medicine and health careNeural Computing and Applications, 2019
- A Deeper Look at Dataset Bias2017
- ImageNet Large Scale Visual Recognition ChallengeInternational Journal of Computer Vision, 2015
- Dental Students’ Ability to Detect and Diagnose Oral Mucosal LesionsJournal of Dental Education, 2015
- A review of risk factors and genetic alterations in head and neck carcinogenesis and implications for current and future approaches to treatmentJournal of Cancer Research and Clinical Oncology, 2009
- Health-related quality of life three years after diagnosis of head and neck cancer?A longitudinal studyHead & Neck, 2001