Feasibility of infrared spectroscopy for discrimination between gallbladder polyp and gallbladder stone using bile juices

• IR spectroscopic discrimination between gallbladder polyp and stone was attempted.
• Bile juices directly dried on Si surface were measured without sample treatments.
• Dissimilar protein structure/phospholipid concentration was source for distinction.
• Cross-validated discrimination accuracy using support vector machine was 78.6%.

This study has attempted infrared (IR) spectroscopic identification of gallbladder (GB) polyp and GB stone using bile juices as specimens. For drop-n-dry measurement of bile juice samples, Si wafer patterned with circular (3 mm in diameter) spots surrounded by hydrophobic coating was prepared to make the formation of sample droplets consistent in shape, and IR spectra of the dried samples on the substrate were acquired. Several noticeable differences in IR spectral features between GB polyp and GB stone samples were observed. First, the intensities of amide I and II bands considerably differed, thereby indicating the differences in the protein composition in both samples. Second, the absorption of COO− stretching bands arising from amino acid-conjugated bile salts, able to form micelles and increase their solubility, was higher for GB polyp samples, equivalently indicating the less chance of GB stone. Third, the intensities of phospholipids bands were smaller in the GB stone samples since the lower phospholipid concentration led to the decrease of cholesterol solubilization and thus increased the chance of GB stone. The first five principal component scores that describe 95.3% of the total spectral variance were fed into support vector machine (SVM) as a classifier. The cross-validated discrimination accuracy was 78.6%. To improve the accuracy further, a sample pretreatment step capable of enhancing the compositional dissimilarity is necessary for some samples that show similar spectral features.