240 — Automated diagnosis of colon cancer using hyperspectral sensing

Beaulieu et al (10.1002/rcs.1897)

Read on 17 April 2018
#cancer  #colon-cancer  #hyperspectral-imaging  #diagnosis  #image-processing  #computer-vision  #surgery  #tumor  #group:JHU  #group:APL 

Tumors have a slightly different composition than healthy tissue: One way in which this manifests most clearly is in the light-scattering patterns of tumorous tissue under wide-spectrum light. Hyperspectral spectroscopy (HSS) is a technology that utilizes this inhomogeneity to great benefit: HSS enables a surgeon to identify the location and bounds of a colorectal tumor by its appearance alone (without ionizing radiation).

Of the patients enrolled in this study at Johns Hopkins Hospital (with technology developed at JHU Applied Physics Laboratory!), HSS techniques greatly improved both sensitivity and specificity of the diagnoses.

Because so much of tumor identification currently relies on visual identification by a trained professional, this technology has the potential to greatly improve patient outcomes in the OR by fully resecting tumors even when the cancerous tissue is not evident in visible light.

I remember rumors I head as an undergraduate at Hopkins that APL was developing a pair of augmented reality glasses that could detect cancer. (This was obviously long before the existence of Google Glass and way before the existence of systems like the HoloLens.) It seems as though that faraway vision may be much closer than we all thought!