Data for OFDVDnet: A sensor fusion approach for video denoising in fluorescence guided surgery
Data files
Abstract
Many applications in machine vision and medical imaging require the capture of images from a scene with very low radiance, which may result in very noisy images and videos. An important example of such an application is the imaging of fluorescently-labeled tissue in fluorescence-guided surgery. Medical imaging systems, especially when intended to be used in surgery, are designed to operate in well-lit environments and use optical filters, time division, or other strategies that allow the simultaneous capture of low radiance fluorescence video and a well-lit visible light video of the scene. This work demonstrates video denoising can be dramatically improved by utilizing deep learning together with motion and textural cues from the noise-free video.
README: Data for OFDVDnet: A Sensor Fusion Approach for Video Denoising in Fluorescence-Guided Surgery
*Equal contribution from Trevor Seets and Wei Lin
https://doi.org/10.5061/dryad.v6wwpzh3w
Description of the data and file structure
This dataset contains 50 videos (about 100 total minutes) of mock surgery on a chicken thigh using a the OnLume Avata fluorescence guided surgery (FGS) imaging system. Each video contains a fluorescence (top right), reference (bottom left), and overlay (top left) view. The fluorescence and reference videos are captured by their corresponding cameras on the system whereas the overlay is a false color overlay of the fluorescence on top of the reference video. Each individual video has resolution of 768x1024 and is captured at 15 fps.
Code/Software
Code is available for simulating noisy video and training the neural networks used to generate the results from the main paper. The readme file within the code explains how to use the code and where the data should be placed within the code file structure.
Methods
The following passage from the main article explains the dataset capture: "We capture data with a transient lighting-enabled, clinical wide-field FGS imaging system (OnLume Surgical, Madison, WI). OnLume’s system uses two camera sensors for both the reference and fluorescent cameras. In a surgical training model, we inject the near-infrared fluorescent agent, ICG, via syringe into the femoral artery of four chicken thighs to simulate vascular surgery. We prepared varying doses of ICG up to the clinical guidelines of 2.5 mg/mL to generate fluorescent videos with visual contrast and low noise that can be treated as ground truth and simulate much lower fluorescence. In future work, we would like to detect markers that have much fewer photons than our captured fluorescent videos. We capture about 100 minutes of simulated surgical footage with a variety of motion such as cutting, pulling, squeezing, injecting, and working with surgical tools. The 100 minutes of footage is broken up into 590 100-frame long videos. The videos are captured at 15 frames per second at a resolution of 768 × 1024."