Abstract
Hyperspectral imaging is an emerging optical modality in oncology that enables non-invasive, label-free tissue characterization by capturing spatially resolved spectral information across a broad wavelength range. Malignant transformation induces biochemical and microstructural changes that alter tissue optical properties, which can be detected through hyperspectral imaging. This chapter presents the physical principles of hyperspectral imaging for cancer diagnosis and reviews computational analysis approaches, with a particular focus on deep learning–based methods for interpreting high-dimensional spectral–spatial data. Applications of hyperspectral imaging are discussed across major cancer types, including brain, breast, colorectal, head and neck, lung, and skin cancers. Current challenges and future research directions are outlined, highlighting the potential of deep learning–enhanced hyperspectral imaging to support precision oncology and image-guided cancer care.

This publication has 43 references indexed in Scilit: