This project aims to enhance the process of hepatic minimally invasive surgery. Right now, different imaging modalities are used to detect the
motion and location of the liver. Of these imaging modalities are MRI, Ultrasound (US) and CT. However, these imaging modalities have complications/restrictions such as low frame rate or low compatibility with ferrous materials (MRI), low spatial resolution (US) or expose patients and physicians to radiation (CT).
Thus, we attempt to use IR as a surrogate signal to avoid the need for these imaging modalities and to overcome these complications. This study aims to use Infrared (IR) imaging, as a surrogate signal, to identify the respiratory-induced motion (RIM) of the liver. The regions of interest (ROIs) will be the nose and the mouth, which are away from the operating area in the abdomen and so, away from the operating region of the operator. This should increase the reliability of the motion tracking of the liver due to having the ROI away from the physician's workspace avoiding external disturbances which cause noise in the surrogate signal. IR camera has been used in multiple studies to extract the breathing rate (BR) of patients and attempt non-contact spirometry. However, these studies haven't experimented further to see if this can compensate for the internal motion of the liver. The plan is to use IR as the surrogate signal as an input for learning-based models.
We aim to extract features from these IR images that will be used in different correspondence models to estimate the induced internal liver motion. Additionally, next to testing multiple models, we want to attempt to increase the complexity of the widely used regression models using neural networks (NN) to estimate the motion of the liver. This study will be validated by experimenting on human subjects and/or lab phantoms. A final comparative study can be created to see if IR surrogate signals can be self-sufficient or if the addition of another surrogate signal would increase the accuracy of the results.