Related Experiment Video
Updated: Jul 26, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.6K
Deep structured learning with vision intelligence for oral carcinoma lesion segmentation and classification using
Ahmad A Alzahrani1, Jamal Alsamri2, Mashael Maashi3
1Department of Computer Science and Artificial Intelligence, College of Computing, Umm-AlQura University, Mecca, Saudi Arabia.
Scientific Reports
|February 24, 2025
Summary
A new machine learning model, DSLVI-OCLSC, significantly improves early detection of oral carcinoma (OC) using medical images. This approach enhances classification and segmentation, aiming to reduce mortality and improve patient outcomes.
Area of Science:
- Oncology
- Medical Imaging
- Artificial Intelligence
Background:
- Oral carcinoma (OC) poses a significant global health challenge, particularly in low-to-middle-income countries, due to late diagnosis and inadequate treatment.
- Early-stage analysis is crucial for effective treatment, prediction, and survival, yet remains a challenge despite advancements in molecular diagnostics.
- Current precision medicine approaches for OC patients face limitations in timely and accurate diagnosis.
Purpose of the Study:
- To introduce a novel Deep Structured Learning with Vision Intelligence for Oral Carcinoma Lesion Segmentation and Classification (DSLVI-OCLSC) model.
- To enhance the classification and recognition of oral carcinoma using medical imaging.
- To improve early detection rates and reduce cancer-specific mortality.
Main Methods:
- The DSLVI-OCLSC model employs wiener filtering (WF) for noise reduction in medical images.
- ShuffleNetV2 is utilized for extracting high-level deep features, while a convolutional bidirectional long short-term memory network with a multi-head attention mechanism (MA-CNN-BiLSTM) handles recognition.
- Unet3+ is used for segmenting abnormal regions, and the sine cosine algorithm (SCA) optimizes hyperparameters.
Main Results:
- The DSLVI-OCLSC model demonstrated superior performance in classifying and segmenting oral carcinoma lesions.
- Experimental simulations on an OC image dataset confirmed the enhanced capabilities of the proposed method.
- The model achieved a high accuracy of 98.47%, outperforming existing approaches.
Conclusions:
- The DSLVI-OCLSC model offers a promising advancement in the early detection and diagnosis of oral carcinoma.
- The integration of deep learning techniques, including attention mechanisms and advanced segmentation, significantly improves diagnostic accuracy.
- This AI-driven approach has the potential to reduce oral carcinoma mortality and improve patient prognoses.
