Analysis of Colour Space Channel Sensitivity for Bilirubin Detection

Colour space channel sensitivity analysis reveals insights into the spectral behavior of bilirubin samples and their response to different wavelengths. The study confirms that the blue wavelength is the most sensitive to changes in bilirubin concentration, followed by the green and red wavelengths. Additionally, the study explores the impact of various biological and external factors on bilirubin measurement, such as tissue thickness, light scattering ratios, white balance, ISO settings, illumination tones, and light intensities.

The study also delves into the effects of different white balance corrections on bilirubin level prediction, highlighting the importance of accurate color representation. Various color spaces, including RGB, CMY(K), L*a*b*, HSV, YCbCr, and LUV, are evaluated for their correlation with bilirubin concentration, with the results showing varying levels of sensitivity and accuracy.

Furthermore, the study assesses the performance of machine learning models in classifying jaundice based on tissue phantom images. The SVM model emerges as the most accurate in binary classification, while LightGBM, RF, and KNN also demonstrate strong performance. In regression tasks, the models show improved accuracy with the inclusion of additional features, with LightGBM, SVM, and RF models performing the best in predicting bilirubin concentrations.

Overall, the study provides valuable insights into color space channel sensitivity analysis, the impact of biological and external factors on bilirubin measurement, white balance corrections, color space conversions, and machine learning model performance in bilirubin classification and regression tasks.