Combination of pathological and spectroscopic characterization to promote diagnosis of retinal pigment epithelium-Bruch’s membrane complex in a diabetic rat model

Abstract

Diabetic retinopathy (DR) is a common condition of diabetes, and approaches to detecting early DR using the unique characteristics of the retinal pigment epithelium-Bruch’s membrane complex (RBC) have increasingly attracted attention. A diabetic model was established in Sprague-Dawley rats via streptozocin (STZ) injection for 1 (DM1) and 6 months (DM6), confirmed by weekly blood glucose measurement. Serum and retinal tissue-based advanced glycation endproducts (AGE) levels significantly elevated in diabetic rats, and RBC was evaluated by transmission electron microscopy and Raman spectroscopy. The results showed that whole Raman spectra and all marked band intensities could respectively achieve almost equal and accurate discrimination of all animal groups, along with the determination of important molecules from the band data. Further quantitative analyses indicated series of metabolic disturbance due to hyperglycemia were involved while the body self-regulation mechanism still played a role with different effects during the disease progression. Given this, Raman spectroscopy can reliably distinguish the early characterization of DR in addition to providing intrinsic key molecules that is sensitive to identify the early disease progression.

Differentiation and prognostic stratification of acute myeloid leukemia by serum‐based spectroscopy coupling with metabolic fingerprints

“…We implemented classification and regression tree (CRT) algorithm that is a commonly employed statistical classifier using the concept of information entropy. Details have been described in our previous work 25 . The intensity data were used to construct pair‐wise models by CRT algorithm, and the classification result in each model presented the top 10 bands with the most importance which were ranked from post‐hoc analysis (Figure 2B).…”

“…Details have been described in our previous work. 25 The intensity data were used to construct pair-wise models by CRT algorithm, and the classification result in each model presented the top 10 bands with the most importance which were ranked from post-hoc analysis (Figure 2B). Notably, the post-hoc list returned almost all important bands shared by models using control as reference.…”