TY - GEN
T1 - Cross validating hyperspectral with Ultrasound-based skin thickness estimation
AU - Vyas, Saurabh
AU - Meyerle, Jon
AU - Burlina, Philippe
N1 - Publisher Copyright:
© 2014 IEEE.
PY - 2014/6/28
Y1 - 2014/6/28
N2 - Our work is focused on the development of non-invasive methods to estimate skin constitutive elements. Such methods can play an important clinical and scientific role in detecting the early onset of skin tumors. Given current statistics by the American Academy of Dermatology suggesting that more than 10 people die each hour worldwide due to skin related conditions, this has potentially high impact on the delivery of skin cancer diagnostics, and patient mortality and morbidity. It can also serve as a valuable tool for research in cosmetology and pharmaceuticals in general. We combine a physics-based model of human skin with machine learning and hyperspectral imaging to non-invasively estimate physiological skin parameters, including melanosomes, collagen, oxygen saturation, blood volume, and skin thickness. While some prior work has been done in this regard, no validation against ground truth has occurred whatsoever. In this specific study we develop a protocol to validate our methodology for estimating one of these skin parameters, skin thickness, using a dataset of 48 hyperspectral signatures obtained in vivo, and cross-validate our depth estimates with a gold standard obtained via Ultrasound. Relative to this gold standard, we find promising mean absolute errors of less than 0.1 mm for skin thickness estimation.
AB - Our work is focused on the development of non-invasive methods to estimate skin constitutive elements. Such methods can play an important clinical and scientific role in detecting the early onset of skin tumors. Given current statistics by the American Academy of Dermatology suggesting that more than 10 people die each hour worldwide due to skin related conditions, this has potentially high impact on the delivery of skin cancer diagnostics, and patient mortality and morbidity. It can also serve as a valuable tool for research in cosmetology and pharmaceuticals in general. We combine a physics-based model of human skin with machine learning and hyperspectral imaging to non-invasively estimate physiological skin parameters, including melanosomes, collagen, oxygen saturation, blood volume, and skin thickness. While some prior work has been done in this regard, no validation against ground truth has occurred whatsoever. In this specific study we develop a protocol to validate our methodology for estimating one of these skin parameters, skin thickness, using a dataset of 48 hyperspectral signatures obtained in vivo, and cross-validate our depth estimates with a gold standard obtained via Ultrasound. Relative to this gold standard, we find promising mean absolute errors of less than 0.1 mm for skin thickness estimation.
KW - Hyperspectral
KW - Ultrasound
KW - skin physiological parameter estimation
UR - http://www.scopus.com/inward/record.url?scp=85038594217&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85038594217&partnerID=8YFLogxK
U2 - 10.1109/WHISPERS.2014.8077565
DO - 10.1109/WHISPERS.2014.8077565
M3 - Conference contribution
AN - SCOPUS:85038594217
T3 - Workshop on Hyperspectral Image and Signal Processing, Evolution in Remote Sensing
BT - 2014 6th Workshop on Hyperspectral Image and Signal Processing
PB - IEEE Computer Society
T2 - 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014
Y2 - 24 June 2014 through 27 June 2014
ER -