Polarimetric Imaging of Peripheral Nerves: An Intraoperative Aid

Medical Imaging Biophysics Life Sciences Bioengineering Medicine
polarimetric imaging peripheral nerves intraoperative aid rotating crossed polarization imaging nerve identification real-time processing portable system

Polarimetric Imaging Technology Assisting Intraoperative Peripheral Nerve Identification: A Cutting-Edge Study

Peripheral nerves play a crucial role in the sensory and control networks of the human body, and their integrity and proper functioning are essential to our quality of life. Unfortunately, accidental peripheral nerve injuries are not uncommon during surgeries. Such injuries not only lead to functional impairment and pain but also result in unfavorable surgical outcomes. In certain anatomically complex regions (such as the hands, wrists, and neck), peripheral nerves are densely distributed among other tissues, making it challenging to distinguish nerves from surrounding structures during surgical procedures. This increases the risk of nerve damage. Currently, surgeons primarily rely on preoperative imaging techniques (e.g., MRI and ultrasound) and their clinical experience to avoid damaging peripheral nerves during surgery. However, these methods have limitations when identifying small nerves. Additionally, preoperative imaging is static and lacks real-time support for intraoperative decision-making. Thus, developing an intuitive, non-invasive, real-time method to enhance nerve identification is of paramount importance, as it can significantly improve surgical outcomes and reduce the risk of nerve injuries.

To address this issue, a research team from the University of Rochester, including Haolin Liao, David J. Mitten, and Wayne H. Knox, proposed an innovative method for intraoperative peripheral nerve identification. Their research findings were published in the February 2025 issue of Biomedical Optics Express (Vol. 16, No. 2). The study showcases a real-time device leveraging a rotating crossed polarization imaging (RXPI) system. By combining polarized light with advanced lock-in amplification data processing, the system achieves automated nerve identification during surgery. Below is a comprehensive analysis of their research.

Research Process and Experimental Design

1. Research Background and Sample Selection: Chicken Thigh Model

The research team initially selected the sciatic nerve of fresh chicken thighs as their experimental model. The chicken thigh model is widely used for microsurgery training due to its accessibility, cost-effectiveness, and ease of handling. The researchers carefully dissected the thighs to expose the sciatic nerve and its surrounding arteries, veins, muscles, and fat tissues. This provided the theoretical foundation for verifying differences among biological tissues using polarimetric imaging. The chicken sciatic nerve appears as a slightly translucent yellow bundle, with a color similar to surrounding fat tissue and a close spatial distribution—making it an ideal subject for studying enhanced nerve identification.

2. Multispectral Crossed Polarization Imaging (XPI)

The researchers constructed a multispectral crossed polarimetry imaging system (XPI) that utilized linear polarizers (extinction ratio 1000:1) arranged in a crossed linear polarization configuration. The experimental procedure included:

  • Establishing a Crossed Polarization Setup: A polarization state generator (PSG) and a polarization state analyzer (PSA) were orthogonally aligned.
  • Light Source: A white LED was used, paired with a linear variable bandpass filter to explore optimal spectral wavelengths.
  • Image Capture and Analysis: By comparing images taken under different wavelengths, researchers found that nerve tissues exhibited strong polarization signal dependency within the 545 nm to 620 nm wavelength range.
  • Data Normalization: Using min-max normalization to process image brightness minimized the effects of illumination differences across spectral bands.

Through difference-calculation imaging (subtracting pixel values of XPI images at different angles), the contrast between nerves and other tissues was further enhanced, laying a foundation for the next step—rotating crossed polarization imaging.

3. Rotating Crossed Polarization Imaging System (RXPI)

To enable omnidirectional nerve identification, the research team designed a motor-driven RXPI system. The system included:

  • Key Components: A pair of orthogonal polarizers mounted on a 3D-printed rotating platform propelled by gears.
  • Rotational Characteristics: The motor rotated at a constant speed of 160 rpm, while a camera recorded video at 240 frames per second, capturing polarization signal variations of nerve tissues at different angles.
  • Data Acquisition: Rotational intensity signals of five different tissue types were analyzed, revealing that nerve signals were prominently sinusoidal and had greater periodic fluctuations compared to veins, arteries, and muscle tissues.

Data Processing: Lock-In Amplification for Enhanced Nerve Detection

Working Principle of Lock-In Amplification

Lock-in amplification (LIA) is a common signal processing technique for extracting periodic signals from noisy backgrounds. Leveraging the predictable periodic variations in nerve polarization signals, the research team used LIA to enhance nerve signal detection. Key steps include:

  1. Reference Signal Generation: A sinusoidal reference signal, matching the frequency of nerve signals, was generated according to the RXPI system’s rotation speed and camera frame rate.
  2. Signal Mixing and AC Value Calculation: Pixel intensity signals were multiplied by the reference signal, and the AC value (indicating the strength of periodic signal variation) was calculated after subtracting the DC (direct current) offset.
  3. Pixel-Wise Processing: Each pixel underwent individual lock-in processing to create an AC Value Map, where high AC values corresponded to nerve tissues.

The experiments showed that nerve tissues had significantly higher AC values compared to other tissues, highlighting the effectiveness of the lock-in amplification method for classification. Receiver operating characteristic (ROC) curve analysis demonstrated an area under the curve (AUC) of 93%, confirming the classifier’s accuracy in chicken thigh models.

Key Results and Findings

  1. Enhanced Nerve Contrast: The AC Value Map highlighted nerve tissues more distinctly in grayscale images, confirming the RXPI system’s high sensitivity to nerve polarization characteristics.
  2. Real-Time Nerve Masking: The team developed a portable prototype device weighing only 525 g, compatible with smartphone cameras. This system enabled real-time data acquisition and processing via lock-in amplification. The AC Value Map refreshed at one frame per second, providing reliable nerve masking assistance during surgical simulations.
  3. Validation in Cadaver Studies: RXPI system tests on fresh-frozen human cadaver forearms successfully enhanced nerve contrasts. Notably, shorter wavelengths (e.g., 460 nm) were more effective for human tissues, suggesting the system’s adaptability across different biological structures.

Significance and Future Prospects

This study introduces a portable, low-cost, automated intraoperative peripheral nerve imaging method with significant scientific and clinical implications:

  1. Scientific Value: By combining RXPI and lock-in amplification, this research explores the polarization characteristics of nerves, offering new insights for biomedical optical techniques.
  2. Clinical Viability: The system’s portability and real-time capabilities make it promising for preventing nerve injuries in surgery and training novice surgeons. It also holds potential for integration with smart devices (e.g., wearable glasses or surgical navigation systems).
  3. Innovation and Practicality: The self-designed RXPI system, paired with innovative data processing, achieves automatic, real-time nerve identification.

Future research directions include live human nerve experiments, further device and software optimization, and incorporating deep learning algorithms for enhanced tissue recognition and classification. Additionally, a direct projection system could be developed to overlay enhanced nerve images onto surgical fields, improving surgeons’ efficiency.

This groundbreaking study not only advances technology in the field of biomedical optics but also opens new pathways in designing surgical assistive tools, greatly enhancing their clinical application potential.