Starting with some specific types of cancers, this project will try to generalize the methodology to discriminate between healthy and malignant tissues in real-time during surgical procedures. Using the hyperspectral signatures of the healthy tissues and the same tissues affected by cancer, a mathematical model of how cancer affects to the hyperspectral signature will be derived. The research will start with the challenging task of brain cancer detection. A precise resection of the gliomas will minimize the negative effect of removing brain cells while assuring an effective tumour resection. The second type of tumours to be analysed will be the lung and breast cancers as they represent the two most common cancers in the world. Based on the experience gained during the evolution of the project and guided by the oncologist expertise, many other types of cancer out
from the more than 200 that affect human beings will be studied. As cancer supposes a change in the cellular physiology, it should be detected as a change in the hyper-spectral signature. This project will try to determine if there is a certain pattern that could be identified as a cancer hyperspectral signature. Although previous works demonstrates that hyperspectral imaging can be used for certain cancer detection in animals, no application to human beings in real-time surgery has been found. This project will develop an experimental intraoperative setup based on non-invasive hyperspectral cameras connected to a platform running a set of algorithms capable of discriminate between healthy or pathological tissues. This information will be provided, through different display devices to the surgeon, overlapping normal viewing images with simulated colours that will indicate the cancer probability of the tissue presently exposed during every instant of the surgical procedure. A high-efficiency hardware/software prototype will be developed with the aim of recognising cancer tissues on real time.