Abstract
AIM: To identify lesional and nonlesional tissues from early gastric cancer (EGC) patients by Raman spectroscopy to build a diagnostic model and effectively diagnose EGC. METHOD: Specimens were collected by endoscopic submucosal dissection from 13 patients with EGC, and 55 sets of standard Raman spectral data (each integrated 10 times) were obtained using the fiber optic Raman system; there were 33 sets of lesional tissue data, including 18 sets of high-grade intraepithelial neoplasia (HGIN) data and 15 sets of adenocarcinoma data, and 22 sets of nonlesional tissue data. After the preprocessing steps, the average Raman spectrum was obtained. RESULTS: The nonlesional tissues showed peaks at 891 cm(-1), 1103 cm(-1), 1417 cm(-1), 1206 cm(-1), 1234 cm(-1), 1479 cm(-1), 1560 cm(-1), and 1678 cm(-1). Compared with the peaks corresponding to nonlesional tissues, the peaks of the lesional tissues shifted by different magnitudes, and a new characteristic peak at 1324 cm(-1) was observed. Comparing the peak intensity ratio and the integral energy ratio of the lesional tissues with those of the nonlesional tissues revealed a significant difference between the two groups (independent-samplest-test, P < 0.05). Considering the peak intensity ratio of I1560 cm(-1)/I1103 cm(-1) as a diagnostic indicator, the accuracy, sensitivity, and specificity of diagnosing EGC were 98.8%, 93.9%, and 91.9%, respectively. Considering the integral energy ratio (noncontinuous frequency band and continuous frequency band) as a diagnostic indicator, the accuracy, sensitivity, and specificity of diagnosing EGC were 99.2-99.6%, 93.9-97.0%, and 95.5%, respectively. CONCLUSIONS: The integral energy ratio of the Raman spectrum could be considered an effective indicator for the diagnosis of EGC.