Related Experiment Video
Updated: Jul 25, 2025
11:27
Quantitative Multispectral Analysis Following Fluorescent Tissue Transplant for Visualization of Cell Origins, Types, and Interactions
Published on: September 22, 2013
9.4K
Joint Diagnostic Method of Tumor Tissue Based on Hyperspectral Spectral-Spatial Transfer Features
Jian Du1,2, Chenglong Tao1,2, Shuang Xue1,2
1Key Laboratory of Spectral Imaging Technology CAS, Xi'an Institute of Optics and Precision Mechanics, Chinese Academy of Sciences, Xi'an 710119, China.
Diagnostics (Basel, Switzerland)
|June 28, 2023
Summary
A new spectral-spatial transfer feature method enhances tumor tissue diagnosis by combining hyperspectral imaging and pathology data. This approach achieves high accuracy in identifying gastric and thyroid cancers rapidly.
Area of Science:
- Medical imaging
- Computational pathology
- Biomedical engineering
Background:
- Hyperspectral imaging (HSI) shows promise for tumor tissue diagnosis but faces challenges with limited medical data.
- Integrating HSI with conventional pathology data can improve diagnostic accuracy and efficiency.
- Transfer learning models can leverage large conventional datasets to overcome limitations in medical HSI data.
Purpose of the Study:
- To develop a joint diagnostic method using spectral-spatial transfer features for improved clinical application of hyperspectral technology in tumor pathology.
- To explore differences in spectral-spatial transfer features between tumor and normal tissues within the 410-900 nm wavelength range.
- To establish a rapid and accurate pathological diagnosis solution using advanced computational methods.
Main Methods:
- A spectral-spatial transfer convolutional neural network (SST-CNN) was developed, pre-trained on conventional pathology datasets.
- The model was applied to micro-hyperspectral images for tumor classification, simulating clinical diagnosis.
- The method combined micro-hyperspectral imaging with large-scale pathological data for feature extraction.
Main Results:
- The SST-CNN model achieved high classification accuracies: 95.46% for gastric cancer and 95.89% for thyroid cancer.
- The joint diagnostic method significantly outperformed models trained solely on conventional digital pathology or hyperspectral data.
- The interpretation of a data section was completed within 3 minutes, demonstrating rapid diagnostic capability.
Conclusions:
- The developed joint diagnostic method based on SST-CNN offers a novel technical solution for rapid and accurate pathological diagnosis of tumor tissues.
- This approach effectively integrates spectral and spatial information from hyperspectral and conventional pathology data.
- The study provides a solid theoretical foundation for hyperspectral pathological diagnosis, addressing feature correlation and efficient transformation.
