In vivo detection of epithelial neoplasia in the stomach using image-guided Raman endoscopy.

This study explores the utility of a novel image-guided Raman endoscopy technique for in vivo distinction of gastric cancer from normal tissue during clinical gastroscopy. The rapid-acquisition fiber-optic Raman endoscopy system developed was employed for in vivo gastric tissue Raman measurements at 785 nm laser excitation. A total of 1063 in vivo Raman spectra were acquired from 238 tissue sites of 67 gastric patients who underwent endoscopic ultrasound staging procedure, in which 934 Raman spectra were from 121 normal tissue sites whereas 129 Raman spectra were from 117 neoplastic gastric tissue sites. Gastric Raman spectra were fitted and reconstructed by using a linear combination of the eight basis reference spectra from the biochemicals (i.e., actin, albumin, collagen, DNA, histones, pepsinogen, phospholipids and triolein) in gastric tissue and also compared with the in vivo gastric Raman spectra measured. The resulting fit coefficients were further utilized through recursive partitioning techniques to develop diagnostic algorithms for gastric cancer diagnosis. High-quality in vivo Raman spectra in the range of 800-1800 cm(-1) can be acquired from gastric mucosa within 0.5s. The fit coefficients from albumin, nucleic acid, phospholipids and histones were found to be the most significant features for construction of the diagnostic model, giving rise to an overall accuracy of 93.7% (i.e., sensitivity of 94.0% (110/117) and specificity of 93.4% (113/121)) for in vivo discrimination of cancerous tissue from normal gastric tissue after the leave-one tissue site-out, cross-validation technique. This work demonstrates for the first time that image-guided Raman endoscopy technique has promising potential for the non-invasive, in vivo diagnosis and detection of gastric cancer at the molecular level.