Early diagnosis of ulcerative colitis based on urine SERS and multivariate analysis

Abstract

Ulcerative colitis (UC) is a type of inflammatory bowel disease (IBD), which can be diagnosed by a combination of invasive endoscopy, clinical examination and histopathology. However, Crohn's disease (CD), as another subtype of IBD, exhibits similar clinical and pathological manifestations to UC. This usually leads to delayed diagnosis of UC and increases the possibility of conversion to colorectal cancer. Moreover, it is difficult to distinguish patients with UC from healthy controls (HC) in early diagnosis. Therefore, it is very important to accurately distinguish IBD subtypes and early identification of UC patients and HC in clinical practice. In this work, surface-enhanced Raman spectroscopy (SERS) was used to measure urine samples from 166 IBD patients (including 83 UC and 83 CD) and 83 HC, totaling 249 samples. It shows that there are significant differences in the intensities at several main Raman peaks between UC and CD, as well as between UC and HC. Then, classification models based on principal component analysis (PCA) were constructed for UC/CD and UC/HC to reduce the dimensionality of the dataset, followed by analysis of the reduced-dimension samples using the constructed support vector machine (SVM) models. After the leave-one-out cross-validation (LOOCV), the UC/CD classification model achieved an accuracy of 72.89 %, sensitivity of 67.47 %, and specificity of 78.31 %, while the UC/HC classification model achieved 82.53 %, 90.36 %, and 74.70 % respectively for these metrics. Therefore, this indicates that SERS, based on urine samples combined with PCA and SVM analysis techniques, can effectively identify IBD subtypes and accurately diagnose UC patients from HC patients, greatly facilitating the clinical diagnosis process. The application of this method has great potential to achieve non-invasive early diagnosis of UC, thereby providing more timely and effective treatment strategies for patients.