Related Experiment Video
Updated: Jun 11, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
6.7K
AutoLNMNet: Automated Network for Estimating Lymph-Node Metastasis in EGC Using a Pyramid Vision Transformer and Data
Lin Gao1, Wenju Liu2, Bingzi Kang1
1School of Science, Jimei University, Xiamen, China.
Microscopy Research and Technique
|October 1, 2024
Summary
A new AI model, AutoLNMNet, accurately detects lymph node metastasis in early gastric cancer (EGC) using multiphoton microscopy and collagen fiber analysis. This tool aids in precise diagnosis and treatment decisions for EGC.
Area of Science:
- Oncology
- Medical Imaging
- Artificial Intelligence
Background:
- Lymph-node status is critical for early gastric cancer (EGC) treatment decisions.
- Endoscopic submucosal dissection is standard for EGC, but accurate staging remains challenging.
- Multiphoton microscopy offers insights into tissue morphology, specifically collagen fibers.
Purpose of the Study:
- To evaluate deep learning models for assessing lymph node metastasis in EGC using collagen fiber morphology.
- To develop and validate a novel AI model, AutoLNMNet, for improved lymph node metastasis detection.
Main Methods:
- Compared four deep learning models (VGG16, ResNet34, MobileNetV2, PVTv2) on preprocessed images.
- Integrated features from the best model (PVTv2) with clinical data to create AutoLNMNet.
- Assessed AutoLNMNet's accuracy and receiver operating characteristics for lymph node metastasis staging (Ly0 and Ly1).
Main Results:
- The PVTv2 model demonstrated strong performance in initial assessments.
- AutoLNMNet achieved a prediction accuracy of 0.92 for both no metastasis (Ly0) and metastasis (Ly1) stages.
- AutoLNMNet showed high receiver operating characteristic values (0.97 for Ly0, 0.97 for Ly1).
Conclusions:
- AutoLNMNet demonstrates high reliability and accuracy in detecting lymph node metastasis in EGC.
- The model offers a valuable tool for enhancing early diagnosis and treatment planning for EGC.
- Collagen fiber analysis via multiphoton microscopy, integrated with AI, shows promise for cancer staging.
