Optical Diagnosis of Liver Cirrhosis and Hepatocellular Carcinoma using Machine Learning-Assisted Serum Raman Spectroscopy
Abstract
This study introduces a simple yet effective method that integrates Raman spectroscopy (RS) with support vector machines (SVM) for the detection and differentiation of hepatocellular carcinoma (HCC), cirrhosis, and healthy individuals serum analysis. RS revealed a prominent collagen marker band at approximately 1246 cm-1, showing significant elevation in HCC and cirrhotic patients compared to healthy controls. Furthermore, serum levels of aromatic amino acids—including tryptophan (757, 878 cm-1), tyrosine (831, 853 cm-1), and phenylalanine (1004 cm-1)—and cholesterol (548, 699 cm-1) were significantly elevated in patients with cirrhosis and HCC compared to the healthy group. In contrast, β-carotene levels (1157, 1527 cm-1) were significantly reduced in both cirrhotic and HCC patients. Binary machine learning classifications (cirrhosis vs. healthy, HCC vs. healthy, and cirrhosis vs. HCC) achieved 83.3–92.0% sensitivity, 89.3–95.3% specificity, and AUC values of 0.929–0.980, validated using 5-fold subject-wise and leave-one-subject-out (LOSO) cross-validations. Moreover, multiclass classification of three groups correctly assigned individual subjects to their respective categories, achieving accuracies of 80.0–89.3% for healthy, 81.3–84.4% for cirrhosis, and 85.3–93.3% for HCC. The integration of RS and machine learning offers a simple, rapid, and cost-effective diagnostic method for the serum-based differentiation and screening of cirrhosis and HCC.