Hyperspectral imaging for tissue classification, a way toward smart laparoscopic colorectal surgery

Area of Science:

• Surgical Oncology
• Medical Imaging
• Biomedical Engineering

Background:

• Laparoscopic surgery is standard for colorectal cancer but lacks tactile feedback.
• Real-time tissue classification can enhance surgical precision and patient outcomes.

Purpose of the Study:

• To evaluate hyperspectral imaging (HSI) for distinguishing tumor from healthy colorectal tissue ex vivo.
• To assess the accuracy of HSI in a clinical setting for colorectal cancer surgery.

Main Methods:

• Hyperspectral imaging (400-1700 nm) was used on colorectal tissue samples from 32 patients.
• Data were classified using quadratic and support vector machine algorithms after feature reduction.
• A 75%/25% train/test split was employed for model validation.

Main Results:

• Overall accuracy for distinguishing three tissue types (fat, healthy wall, tumor) was 0.88 (STD=0.13).
• Mean per-patient accuracy on the test set reached 0.93 (STD=0.12).

Conclusions:

• Hyperspectral imaging demonstrates significant potential for real-time tissue classification in colorectal cancer surgery.
• Further research should focus on intraoperative application and whole specimen imaging.