Computer-aided Veress needle guidance using endoscopic optical coherence tomography and convolutional neural networks

Area of Science:

• Minimally Invasive Surgery
• Medical Imaging Technology
• Surgical Navigation Systems

Background:

• Veress needle insertion for pneumoperitoneum in laparoscopic surgery presents placement challenges.
• Accurate needle tip localization is crucial for patient safety during abdominal procedures.

Purpose of the Study:

• To develop and evaluate a computer-aided endoscopic optical coherence tomography (OCT) system for guiding Veress needle insertion.
• To improve the safety and precision of Veress needle placement in laparoscopic surgery.

Main Methods:

• Development of an endoscopic OCT system integrated with convolutional neural networks (CNNs).
• Testing the system on swine tissue samples (subcutaneous fat, muscle, abdominal space, small intestine) to simulate surgical conditions.
• Utilizing CNNs for automatic tissue classification and Veress needle tip localization based on OCT imaging.

Main Results:

• The system achieved high accuracy in tissue classification (98.53% ± 0.39%).
• Accurate estimation of needle-to-tissue distance with a low average relative error (4.42% ± 0.56%).
• Successful visualization and differentiation of various abdominal tissue layers and spaces.

Conclusions:

• The developed endoscopic OCT system effectively guides Veress needle insertion.
• Computer-aided navigation using OCT and CNNs enhances precision and safety in laparoscopic surgery.
• This technology holds potential for reducing complications associated with Veress needle placement.