Graph Neural Network With Hessian-Based Locally Linear Embedding for Cancer Metastasis Analysis in Lymph Nodes Using DeepLab Segmentation
Breast cancer patients could benefit greatly from an automatic diagnosis of metastases in lymph nodes. In clinical settings, where whole-slide images of histological lymph nodes are involved, it is crucial to identify and detect breast cancer metastases. Patient survival rates increase significantly...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10908385/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Breast cancer patients could benefit greatly from an automatic diagnosis of metastases in lymph nodes. In clinical settings, where whole-slide images of histological lymph nodes are involved, it is crucial to identify and detect breast cancer metastases. Patient survival rates increase significantly with early detection. In order to identify metastases, pathologists are depending more and more on microscopic research. However, the diagnostic process is time-consuming and sometimes results in missed diagnoses. Automated, reliable patient-level classification has the potential to minimize pathologists’ effort while improving diagnosis accuracy. The objective of this work is to propose and standardize a new framework that combines a Graph Neural Network (GNN) with Hessian-based Locally Linear Embedding (HLLE) to better understand cancer metastasis in lymph nodes, while DeepLab is used to effectively segment each location. The proposed framework integrates DeepLab, a highly advanced convolutional neural network for image segmentation, to segment the lymph node regions from high-resolution MRI and PET scans. These segmented regions are then modeled as graphs, where the nodes signify the segmented areas and the edges reflect spatial and functional coupling. To accomplish better preservation of the geometry of high-dimensional data and more effective feature extraction, the HLLE algorithm is applied. These features are then inputted into a GNN, which utilizes the graph structure to identify and analyze metastatic patterns. The method was tested on the CAMELYON17 dataset, which contains of imaging data from patients with dissimilar stages of cancer. Precisely, DeepLab was utilized to segment the lymph nodes, and the segmented regions were then converted into graphs. The meaningful features were obtained utilizing the proposed HLLE, after which the GNN conducted the classification and metastasis analysis. Overall, the precision and reliability of the established segmentation method, DeepLab, followed by HLLE and GNN, proved to be precise in classifying metastatic lymph nodes and their classification. |
|---|---|
| ISSN: | 2169-3536 |