Aaron Sun1, Jung-Man Namgoong, MD2, Peter Kim, MD, PhD3, Jaepyeong Cha, PhD3. 1Princeton University, 2Asan Medical Center, 3Children’s National Health System
Objective: Real-time anatomic display of tissue perfusion, vascularity and ischemia in the tissue of interests during surgery is critical and necessary for improvements in technical and cognitive decisions. Snapshot hyperspectral imaging (HSI) has advantages over conventional multispectral imaging methods because of its fast image acquisition of all spectral bands, making it more useful for real-time surgical application. This technology has not been introduced to laparoscopic surgery. Herein, we demonstrate a working algorithm that differentiates between regions of high and low oxygen saturation providing real-time processing and display using a snapshot HSI.
Description: We incorporated a snapshot hyperspectral imager with a standard RGB endoscope. Light reflecting off from the target tissues enters the snapshot hyperspectral imager through the endoscope and the resulting hyperspectral images are de-mosaic and processed by using a custom program routine to determine relative tissue oxygen saturation levels. In vivo ischemic conditions were created in sections of rat bowels by suture ligation or clamping of the vessels. Tissue perfusion was imaged and monitored over the course of time up to an hour.
Results: The endoscopic snapshot hyperspectral imaging system ( in Figure 1) processed images at a video frame rates of approximately 24 frames per second. Figures 2 and 3 illurtrate representative image frames acquired at the end of each experiment before and after image processing. Figure 2 shows an RGB image of the target bowel tissue regions acquired using the hyperspectral color compsensation from the same hyperspectral imager, while Figure 3 shows the resulting heatmap of the tissue perfusion image where the higher oxygen saturation levels are in yellow and lower levels are in blue.
We then analyzed the relative oxygen saturation levels in both healthy and clamped regions of tissue over the course of one hour (Figure 4). Oxygen levels in the healthy region remained relatively constant over the course of the time period whereas oxygen saturation in the clamped regions dropped more than 20% within 30 minutes.
Conclusions and Future Directions: Using a snapshot hyperspectral endoscopic imaging, we were able to differentiate between ischemic and healthy regions. Additionally, we were able to measure the relative mean oxygen saturation levels in each section of tissue. We are currently refining the algorithm using machine learning for image analysis to improve both accuracy and processing speed. This technique can be useful especially in post-operation recovery without complication. This real-time imaging and display can greatly help perform surgery such as organ transplantation or gastro-intestinal surgery in the future.
Presented at the SAGES 2017 Annual Meeting in Houston, TX.
Abstract ID: 91140
Program Number: ETP860
Presentation Session: Emerging Technology iPoster Session (Non CME)
Presentation Type: Poster