Source multiplexing enhances the number of channels of a multispectral sensor

Katsafadou, Maria; Giardini, Mario

doi:10.1109/embc48229.2022.9871570

Cited by 1 publication

(1 citation statement)

References 5 publications

(6 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One of the hand images and one of the foot images with occluded circulation, randomly chosen amongst the five captured, were used to train a simple pixel classifier. In particular, for each image, the red, green, and blue channel values were extracted as, even in the presence of monochromatic illumination, these carry different spectral information 18 , albeit details of the spectral profiles are difficult to determine at this stage, as we do not know the spectral sensitivity of each camera pixel. Additionally, the average intensity of each channel for all illumination wavelengths was calculated and, from this, normalised channel values were calculated.…”

Section: Discussionmentioning

confidence: 99%

At-home multispectral imager for the monitoring of blood oxygenation in people at risk of diabetic foot ulcers

Shand,

Thomson,

Philip

et al. 2024

Clinical Biophotonics III

View full text Add to dashboard Cite

One in three people living with diabetes are affected by diabetic foot ulcers (DFUs) which, if left unmonitored and untreated, reduce quality of life, and may lead to foot amputation. Regular foot screening is suggested by relevant national guidelines. Yet, compliance with screening is poor, mostly due to the inconvenience of the screening process itself. In this work, we describe a preliminary version of a simple low-cost device intended for self-monitoring of foot circulation, to identify the areas of poor perfusion expected to be at the root of the formation of DFUs. This device is based on multispectral imaging of the feet in the visible and near-infrared ranges. The device is tested on the hands and feet of a single subject, where impairment of circulation has been simulated through a brief ligature of a finger/two toes. Multispectral images are captured, and a simple machine-learning-based classifier correctly identifies areas of low perfusion on the hand and shows promising data on toes. Albeit the device and the classifier are still susceptible of significant improvement, this is indicative of the fact that the multispectral images contain relevant information on tissue perfusion.

show abstract

Section: Discussionmentioning

confidence: 99%