Premium
A pilot study on parallel factor analysis as a diagnostic tool for oral cancer diagnosis: A statistical modeling approach
Author(s) -
Kanniyappan Udayakumar,
Gnanatheepam Einstein,
Subramani Karthikeyan,
Rajendran Mangaiyarkarasi,
Chinnathambi Shanmugavel,
Aruna Prakasarao,
Ganesan Singaravelu
Publication year - 2021
Publication title -
journal of chemometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.47
H-Index - 92
eISSN - 1099-128X
pISSN - 0886-9383
DOI - 10.1002/cem.3315
Subject(s) - flavin adenine dinucleotide , nicotinamide adenine dinucleotide , linear discriminant analysis , cancer , chemistry , pathology , medicine , biochemistry , mathematics , statistics , cofactor , enzyme , nad+ kinase
Excitation‐emission matrix (EEM) has been extensively used as the comprehensive diagnostic tool to extract the biochemical fingerprint of the intrinsic fluorophores in a single scan window. However, there is a gap between the rigorous applications of the statistical tool with respect to discrimination of different stages of the disease which has been the subject for many years. Parallel factor analysis (PARAFAC) is one among the powerful statistical modeling approaches among others. In the present study, a total of 70 EEM matrices of normal, premalignant, and malignant oral tissues were given as a input, and seven intrinsic fluorophores were extracted as “components.” The extracted components were well correlated with respect to the appropriate excitation and emission spectral characteristics of the multiple intrinsic fluorophores such as tryptophan, flavin adenine dinucleotide (FAD), nicotinamide adenine dinucleotide (NADH), collagen‐1, porphyrin, tyrosine, and collagen. Subsequently, the student's t test and linear discriminant analysis (LDA) have been carried out with respect to the fluorescence intensity scores between normal vs. premalignant, normal vs. cancer, and premalignant vs. malignant groups. In normal vs. premalignant, all the seven fluorophores exhibit good statistical accuracy except porphyrin; normal vs. cancer exhibits higher statistical significance for tryptophan, NADH, and FAD than rest of the fluorophores, and premalignant vs. malignant shows proper classification in discriminating FAD, collagen‐1, and collagen. In summary, based on positive predictive value, the normal vs. premalignant exhibits 100% classification than the other two groups. Hence, the PARAFAC analysis could be the alternative and useful diagnostic tool in oral cancer diagnosis.