⚡ Quick Summary
This study introduces a novel artificial intelligence-assisted diagnostic approach utilizing the SegFormer semantic segmentation model for the automatic segmentation of lesions in oral mucosal diseases. The model demonstrated a Dice coefficient of 0.710 and a mean Intersection over Union (mIoU) of 0.786, significantly outperforming traditional methods.
🔍 Key Details
- 📊 Dataset: 838 high-resolution images of oral mucosal diseases
- 🧩 Diseases analyzed: Oral lichen planus, oral leukoplakia, oral submucous fibrosis
- ⚙️ Technology: SegFormer model based on Transformer architecture
- 🏆 Performance metrics: Dice coefficient: 0.710, mIoU: 0.786
🔑 Key Takeaways
- 🤖 AI technology can enhance the accuracy of diagnosing oral mucosal diseases.
- 📸 High-resolution images were used to train the segmentation model.
- 🧑⚕️ Expert annotation was performed at the pixel level by oral specialists.
- 🏆 SegFormer-B2 model outperformed classical models like UNet and DeepLabV3.
- 📈 Quantitative metrics such as Dice coefficient and mIoU were utilized for evaluation.
- 🌟 Visual analysis confirmed the model’s ability to accurately segment lesions.
- 🔍 Clinical application of this model could improve diagnostic efficiency.
- 🌍 Study conducted at the Affiliated Stomatological Hospital of Zhejiang University School of Medicine.
📚 Background
Diagnosing oral mucosal diseases can be quite challenging due to their complex nature and the limitations of traditional object detection methods. The need for a more precise and objective approach has led researchers to explore the potential of deep learning technologies, which can significantly enhance diagnostic accuracy and efficiency.
🗒️ Study
This research focused on developing a high-accuracy diagnostic tool using the SegFormer model, which is based on the Transformer architecture. The study utilized a dataset of 838 high-resolution images of three common oral mucosal diseases, annotated meticulously by specialists using Labelme software. The aim was to create a robust semantic segmentation dataset that could facilitate the automatic segmentation of lesion areas in clinical settings.
📈 Results
The SegFormer-B2 model achieved remarkable results, with a Dice coefficient of 0.710 and a mean Intersection over Union (mIoU) of 0.786. These metrics indicate a high level of accuracy in segmenting lesion areas compared to traditional models like UNet and DeepLabV3, which underscores the effectiveness of the proposed approach.
🌍 Impact and Implications
The findings from this study have significant implications for the field of oral medicine. By providing a reliable and efficient tool for the automatic segmentation of lesions, the SegFormer model can enhance the diagnostic process, leading to better patient outcomes. This technology holds promise for broader applications in clinical settings, potentially transforming how oral mucosal diseases are diagnosed and managed.
🔮 Conclusion
This research highlights the transformative potential of deep learning in the diagnosis of oral mucosal diseases. The SegFormer model not only demonstrates high accuracy in lesion segmentation but also paves the way for future advancements in AI-assisted diagnostics. Continued exploration in this area could lead to significant improvements in clinical practices and patient care.
💬 Your comments
What are your thoughts on the integration of AI in diagnosing oral mucosal diseases? We would love to hear your insights! 💬 Share your comments below or connect with us on social media:
Research and Application of Deep Learning Models with Multi-Scale Feature Fusion for Lesion Segmentation in Oral Mucosal Diseases.
Abstract
Given the complexity of oral mucosal disease diagnosis and the limitations in the precision of traditional object detection methods, this study aims to develop a high-accuracy artificial intelligence-assisted diagnostic approach based on the SegFormer semantic segmentation model. This method is designed to automatically segment lesion areas in white-light images of oral mucosal diseases, providing objective and quantifiable evidence for clinical diagnosis. This study utilized a dataset of oral mucosal diseases provided by the Affiliated Stomatological Hospital of Zhejiang University School of Medicine, comprising 838 high-resolution images of three diseases: oral lichen planus, oral leukoplakia, and oral submucous fibrosis. These images were annotated at the pixel level by oral specialists using Labelme software (v5.5.0) to construct a semantic segmentation dataset. This study designed a SegFormer model based on the Transformer architecture, employed cross-validation to divide training and testing sets, and compared SegFormer models of different capacities with classical segmentation models such as UNet and DeepLabV3. Quantitative metrics including the Dice coefficient and mIoU were evaluated, and a qualitative visual analysis of the segmentation results was performed to comprehensively assess model performance. The SegFormer-B2 model achieved optimal performance on the test set, with a Dice coefficient of 0.710 and mIoU of 0.786, significantly outperforming other comparative algorithms. The visual results demonstrate that this model could accurately segment the lesion areas of three common oral mucosal diseases. The SegFormer model proposed in this study effectively achieves the precise automatic segmentation of three common oral mucosal diseases, providing a reliable auxiliary tool for clinical diagnosis. It shows promising prospects in improving the efficiency and accuracy of oral mucosal disease diagnosis and has potential clinical application value.
Author: [‘Zhang R’, ‘Lu M’, ‘Zhang J’, ‘Chen X’, ‘Zhu F’, ‘Tian X’, ‘Chen Y’, ‘Cao Y’]
Journal: Bioengineering (Basel)
Citation: Zhang R, et al. Research and Application of Deep Learning Models with Multi-Scale Feature Fusion for Lesion Segmentation in Oral Mucosal Diseases. Research and Application of Deep Learning Models with Multi-Scale Feature Fusion for Lesion Segmentation in Oral Mucosal Diseases. 2024; 11:(unknown pages). doi: 10.3390/bioengineering11111107