Recent application of Raman spectroscopy in tumor diagnosis: from conventional methods to artificial intelligence fusion

Abstract Raman spectroscopy, as a label-free optical technology, has widely applied in tumor diagnosis. Relying on the different Raman technologies, conventional diagnostic methods can be used for the diagnosis of benign, malignant and subtypes of tumors. In the past 3 years, in addition to traditional diagnostic methods, the application of artificial intelligence (AI) in various technologies based on Raman technologies has been developing at an incredible speed. Based on this, three technical methods from single spot acquisition (conventional Raman spectroscopy, surface-enhanced Raman spectroscopy) to Raman imaging are respectively introduced and analyzed the diagnosis process of these technical methods. Meanwhile, the emerging AI applications of tumor diagnosis within these methods are highlighted and presented. Finally, the challenges and limitations of existing diagnostic methods, and the prospects of AI-enabled diagnostic methods are presented.