Related Experiment Video
Updated: Jan 6, 2026
11:05
Tissue-simulating Phantoms for Assessing Potential Near-infrared Fluorescence Imaging Applications in Breast Cancer Surgery
Published on: September 19, 2014
12.6K
Neural illumination calibration for surgical workflow-optimized spectral imaging
Alexander Baumann1,2,3, Leonardo Ayala4,5, Alexander Studier-Fischer6,7,8,9
1Siemens AG, Munich, Germany. baumann.alexander@siemens.com.
International Journal of Computer Assisted Radiology and Surgery
|October 7, 2025
Summary
This study introduces an automated hyperspectral imaging (HSI) calibration method for surgery. The learning-based approach ensures accurate surgical scene analysis under changing lights, improving HSI integration.
Area of Science:
- Medical Imaging
- Surgical Technology
- Computer Vision
Background:
- Hyperspectral imaging (HSI) offers significant potential for surgical applications.
- Current HSI cameras face integration challenges due to manual recalibration needs under changing lighting.
Purpose of the Study:
- To develop a learning-based approach for automatic, in-situ recalibration of hyperspectral cameras during surgery.
- To enable seamless integration of HSI into clinical workflows by addressing illumination variability.
Main Methods:
- A novel method predicts white reference images from uncalibrated HSI data.
- Combines real-world data with physics-inspired simulations for diverse training.
- Disentangles illumination variations from tissue characteristics.
Main Results:
- Dynamic lighting significantly impacts surgical analysis; the proposed method overcomes this.
- The approach achieves high accuracy, surpassing manual recalibration and prior methods.
- Demonstrates robust generalization across species, lighting conditions, and tasks.
Conclusions:
- The developed method facilitates automated, rapid illumination calibration for HSI in surgery.
- Enhances the reliability and practicality of spectral imaging in clinical settings.
- Paves the way for wider adoption of HSI in surgical procedures.
