Invasive fungal infection serves as a great threat to human health. Discrimination between fungal and bacterial infections at the earliest stage is vital for effective clinic practice; however, traditional culture-dependent microscopic diagnosis of fungal infection usually requires several days, meanwhile, culture-independent immunological and molecular methods are limited by the detectable type of pathogens and the issues with high false-positive rates. In this study, we proposed a novel culture-independent phenotyping method based on single-cell Raman spectroscopy for the rapid discrimination between fungal and bacterial infections. Three Raman biomarkers, including cytochrome c, peptidoglycan, and nucleic acid, were identified through hierarchical clustering analysis of Raman spectra across 12 types of most common yeast and bacterial pathogens. Compared to those of bacterial pathogens, the single cells of yeast pathogens demonstrated significantly stronger Raman peaks for cytochrome c, but weaker signals for peptidoglycan and nucleic acid. A two-step protocol combining the three biomarkers was established and able to differentiate fungal infections from bacterial infections with an overall accuracy of 94.9%. Our approach was also used to detect ten raw urinary tract infection samples. Successful identification of fungi was achieved within half an hour after sample obtainment. We further demonstrated the accurate fungal species taxonomy achieved with Raman-assisted cell ejection. Our findings demonstrate that Raman-based fungal identification is a novel, facile, reliable, and with a breadth of coverage approach, that has a great potential to be adopted in routine clinical practice to reduce the turn-around time of invasive fungal disease (IFD) diagnostics.

侵袭性真菌感染是对人类健康的极大威胁。在早期阶段区分真菌和细菌感染对于有效的临床实践至关重要；然而，传统的依赖于培养的真菌感染显微诊断通常需要几天时间，同时，不依赖于培养的免疫学和分子学方法受到可检测病原体类型和高假阳性率问题的限制。在这项研究中，我们提出了一种基于单细胞拉曼光谱的新型非培养表型分析方法，用于快速区分真菌和细菌感染。三种拉曼生物标志物，包括细胞色素c、肽聚糖和核酸，是通过对 12 种最常见的酵母和细菌病原体的拉曼光谱进行层次聚类分析来鉴定的。与细菌病原体相比，酵母病原体的单细胞显示出明显更强的细胞色素c拉曼峰, 但肽聚糖和核酸的信号较弱。建立了结合三种生物标志物的两步方案，能够区分真菌感染和细菌感染，总体准确率为 94.9%。我们的方法还用于检测十个原始尿路感染样本。取样后半小时内，真菌鉴定成功。我们进一步证明了通过拉曼辅助细胞喷射实现的准确真菌物种分类。我们的研究结果表明，基于拉曼的真菌鉴定是一种新颖、简便、可靠且覆盖面广的方法，在常规临床实践中具有很大的潜力，可以缩短侵袭性真菌病的周转时间（IFD ）诊断。