Framework for hyperspectral image processing and quantification for cancer detection during animal tumor surgery.

Area of Science:

• Biomedical optics
• Medical imaging
• Cancer diagnostics

Background:

• Hyperspectral imaging (HSI) offers potential for real-time cancer detection in surgery.
• HSI data requires robust processing to distinguish cancerous from normal tissue.
• Accurate tumor margin assessment is critical for effective surgical outcomes.

Purpose of the Study:

• To develop and validate a hyperspectral image processing and quantification framework.
• To enhance the accuracy and speed of cancer detection using HSI.
• To assess the feasibility of HSI for tumor margin assessment in image-guided surgery.

Main Methods:

• Developed a framework including image preprocessing, glare removal, feature extraction, and classification.
• Applied the framework to HSI data (450-900 nm) from mice with head and neck cancer.
• Utilized Fourier coefficients, normalized reflectance, mean, and spectral derivatives for analysis.

Main Results:

• Demonstrated the feasibility of the HSI processing framework for cancer detection in animal models.
• Achieved improved accuracy in differentiating cancerous tissue from normal tissue.
• Showcased high potential for fast image classification, crucial for surgical settings.

Conclusions:

• The proposed HSI framework is effective for cancer detection and tumor margin assessment.
• This approach offers a promising tool for image-guided surgery, especially when rapid assessment is paramount.
• Further application in clinical settings may improve surgical precision and patient outcomes.