Advancing cancer diagnostics with artificial intelligence and spectroscopy: identifying chemical changes associated with breast cancer

• Published in: Expert review of molecular diagnostics Vol. 19; no. 10; pp. 929 - 940
• Main Authors: Talari, Abdullah C.S., Rehman, Shazza, Rehman, Ihtesham U
• Format: Journal Article
• Language: English
• Published: England Taylor & Francis 03.10.2019
• Subjects: artificial intelligence; Breast cancer; principal component and linear discriminant analysis; Raman spectroscopy; Tissue microarray (TMA) biopsies; Tissue microarray (TMA) biopsies; principal component and linear discriminant analysis; Breast cancer; Raman spectroscopy; artificial intelligence
• ISSN: 1473-7159; 1744-8352
• DOI: 10.1080/14737159.2019.1659727

Background: Artificial intelligence (AI) and machine learning (ML) approaches in combination with Raman spectroscopy (RS) to obtain accurate medical diagnosis and decision-making is a way forward for understanding not only the chemical pathway to the progression of disease, but also for tailor-made personalized medicine. These processes remove unwanted affects in the spectra such as noise, fluorescence and normalization, and help in the optimization of spectral data by employing chemometrics. Methods: In this study, breast cancer tissues have been analyzed by RS in conjunction with principal component (PCA) and linear discriminate (LDA) analyses. Tissue microarray (TMA) breast biopsies were investigated using RS and chemometric methods and classified breast biopsies into luminal A, luminal B, HER2, and triple negative subtypes. Results: Supervised and unsupervised algorithms were applied on biopsy data to explore intra and inter data set biochemical changes associated with lipids, collagen, and nucleic acid content. LDA predicted specificity accuracy of luminal A, luminal B, HER2, and triple negative subtypes were 70%, 100%, 90%, and 96.7%, respectively. Conclusion: It is envisaged that a combination of RS with AI and ML may create a precise and accurate real-time methodology for cancer diagnosis and monitoring.