Skin Hydration And Transepidermal Water Loss Measurement Using Vis/nir Spectroscopy And Feed-forward Backpropagation Neural Network
Skin hydration and transepidermal water loss (TEWL) are commonly measured using electrical method. However, electrical method can be affected by numerous factors such as the physical condition of the probe due to corrosion and oxidation, ambient temperature and humidity, and the underlying ions conc...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2022
|
| Subjects: | |
| Online Access: | http://eprints.usm.my/60074/1/TAN%20CHUN%20HO%20-%20TESIS24.pdf http://eprints.usm.my/60074/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Skin hydration and transepidermal water loss (TEWL) are commonly measured using electrical method. However, electrical method can be affected by numerous factors such as the physical condition of the probe due to corrosion and oxidation, ambient temperature and humidity, and the underlying ions concentration under the skin. Thus, this study proposed the use of visible/near-infrared (VIS/NIR) spectroscopy, specifically at wavelengths 698 nm, 970 nm, 1200 nm, 1450 nm, and 1950 nm to predict the skin hydration and TEWL. In this study, the prediction models were built using simple linear regression, multiple regression analysis, and feed-forward backpropagation neural network (FFBPNN) with gradient descent algorithm. The FFBPNN consists of an input layer, three hidden layers with five hidden nodes per each hidden layer, and an output layer. Twenty four participants were recruited for this study; their skin hydration, TEWL, and absorbance spectra were collected from the participants’ palms and arms. The participants were comprised of various skin colours, ethnics, and genders, with age ranges from 22 to 32 years old. The participants were selected at random with the main criteria that the healthy subjects were without any skin diseases. The experiment was carried out in an air-conditioned room under controlled temperature. The temperature of the air conditioner setting was set at 24 °C and the settings remain constant throughout the data collection. The data were then randomly sorted into three datasets; the first set was used for the training of FFBPNN, while the second and third dataset were used for the model selection and validation test respectively. |
|---|