Automated System Enables Real-Time "Molecular Pathology" During Cancer Surgery

Positive surgical margins occur in up to 40% of cases, increasing the risk of recurrence and impairing postoperative function. Surgeons often rely on frozen-section analysis, but this approach is slow, subjective, and difficult to apply comprehensively during an operation. Now, a new technology aims to solve this problem by enabling real-time detection of tumor aggressiveness directly at the surgical site.

Researchers at the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences (SIAT, Guangdong, China) and Fudan University (Shanghai, China) have developed a label-free intraoperative navigation system powered by surface-enhanced Raman scattering. The platform integrates a microfluidic sampling pen, a nanoimprinted SERS array, and an AI-driven spectral interpretation module to measure tissue acidity and PSA activity in situ. The pen gently extracts biomarker-containing droplets from tissue surfaces within six seconds per point and transfers them to a uniform SERS array functionalized with pH- and PSA-sensitive Raman reporters. A custom two-dimensional deep learning model then processes the spectra to deliver quantitative malignancy readings in under two minutes.

In a clinical trial of 144 prostate cancer patients, the system demonstrated strong diagnostic performance with an AUC of 0.890 for identifying high-grade tumors, exceeding the accuracy of conventional intraoperative tools such as frozen-section pathology. The automated nature of the workflow reduces operator dependence and offers surgeons timely, objective information on tumor biology. By generating a malignancy “heat map” of the resection area, the system supports more precise surgical decision-making, helping clinicians remove aggressive tumor tissue while preserving nerves and other structures essential for continence and sexual function.

This technology, presented in Nature Biomedical Engineering, represents a major step toward real-time molecular pathology during surgery. Its ability to provide spatially resolved, quantitative tumor profiling could guide broader applications in precision oncology and optimize outcomes for patients undergoing complex cancer resections. Future development will focus on scaling the platform to additional tumor types and validating its performance across more surgical centers.

Related Links:
SIAT
Fudan University