Related Experiment Video
Updated: Feb 17, 2026
Industrialized, Artificial Intelligence-guided Laser Microdissection for Microscaled Proteomic Analysis of the Tumor Microenvironment
Published on: June 3, 2022
Image-guided thoracoscopic segmentectomy via single-shot statistical gating of blood-scattered laser speckle
Haoji Ma1, Hua Liu2,3, Zichen Wang1
1School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, 200240, China.
Abstract:
Accurate identification of target lung segments during thoracoscopic surgery is critical for successful lung cancer resections but remains challenging with conventional thoracoscopic imaging techniques. We present a real-time statistical gating method that leverages the single-shot laser speckle pattern generated by an 840 nm laser to isolate the blood-scattered speckle component in lung tissue. This enables background-free measurement of hemoglobin's absorption, thereby markedly improving the sensitivity of tissue oxygen saturation detection. By exploiting differences in speckle decorrelation time, our approach reconstructs blood-scattered intensity images from single-frame thoracoscopic captures in real-time. Clinical trials demonstrated a 2.05-fold increase in boundary slope steepness and a 220% improvement in the mean absolute derivative metric, enabling precise differentiation of the segment's boundary during the inflation-deflation procedure in lung segmentectomy. For novice surgeons, manual segmentation accuracy improved from 0.78 to 0.92 (standard segmentectomy) and from 0.82 to 0.89 (rapid segmentectomy) in terms of DICE coefficients. Compatible with standard thoracoscopic systems, our method offers real-time, high-sensitivity visualization of blood absorption dynamics, enhancing surgical precision and reducing operative time. As a computational imaging method, the proposed statistical gating method can be seamlessly integrated into existing thoracoscopic systems by adding near-infrared laser illumination and an embedded GPU core. This highlights its potential for revolutionizing lung cancer surgeries.
Insights
This study introduces a real-time statistical gating method using near-infrared laser speckle imaging to precisely identify lung segments during surgery. The technique improves visualization of blood absorption, enhancing surgical accuracy and outcomes for lung cancer resections.
Area of Science:
- Medical Imaging
- Computational Imaging
- Surgical Technology
Background:
- Accurate lung segment identification is crucial for lung cancer surgery but challenging with current thoracoscopic imaging.
- Conventional methods struggle with precise visualization of tissue boundaries and oxygenation status.
Purpose of the Study:
- To develop and validate a real-time statistical gating method for enhanced lung segment identification during thoracoscopic surgery.
- To improve the sensitivity and accuracy of tissue oxygen saturation detection for better surgical guidance.
Main Methods:
- Utilized a near-infrared (840 nm) laser to generate single-shot laser speckle patterns.
- Developed a statistical gating technique to isolate blood-scattered speckle components and measure hemoglobin absorption.
- Reconstructed blood-scattered intensity images in real-time by exploiting speckle decorrelation time differences.
Main Results:
- Demonstrated a 2.05-fold increase in boundary slope steepness and a 220% improvement in mean absolute derivative metric.
- Significantly improved manual segmentation accuracy for novice surgeons (DICE coefficients from 0.78 to 0.92).
- Enabled precise differentiation of lung segment boundaries during inflation-deflation procedures.
Conclusions:
- The statistical gating method offers real-time, high-sensitivity visualization of blood absorption dynamics, enhancing surgical precision.
- This computational imaging approach is compatible with standard thoracoscopic systems and has the potential to revolutionize lung cancer surgery.
- The technique improves operative time and accuracy, particularly for novice surgeons, by providing clear visualization of target lung segments.
