Diagnosis of oral cancer by light-induced auto¯uorescence spectroscopy using double excitation wavelengths Chih-Yu Wang a , Huihua Kenny Chiang a, *, Chin-Tin Chen b , Chun-Pin Chiang c , Ying-Shiung Kuo c , Song-Nan Chow b a Institute of Biomedical Engineering, National Yang-Ming University 155, Section 2, Li-Nung Street, Taipei, Taiwan b Laser Medicine Research Center, National Taiwan University, Taipei, Taiwan c School of Dentistry, National Taiwan University, Taipei, Taiwan Received 22 September 1998; accepted 14 November 1998 Abstract A cancer diagnostic algorithm, light-induced auto¯uorescence spectroscopy using double excitation wavelengths, was employed for distinguishing between cancerous and normal oral mucosa. For emission spectra at the shorter excitation wavelengths (280, 290, and 300 nm), the ratio between the area under 325±335 nm and the area under 465±475 nm was calculated. In the same way, for emission spectra at the longer excitation wavelengths (320, 330, and 340 nm), the ratio between the area under 375±385 nm and the area under 465±475 nm was calculated. Receiver operating characteristic curves were used to evaluate the performance of algor- ithms using single and the double (by combining shorter and longer) excitation wavelengths. The results showed that better per- formance, up to sensitivity 81.25%, speci®city 93.75%, and positive predictive value 92.86%, could be achieved by using the double excitation wavelengths. The present study can be useful as a basis for further investigation on in vivo auto¯uorescence measurement and analysis using double excitation wavelength. # 1999 Elsevier Science Ltd. All rights reserved. Keywords: Auto¯uorescence spectroscopy; Double excitation wavelengths; Oral cancer 1. Introduction Oral cancer has become one of the leading malig- nancies in many Asian countries [1], where harmful habits such as smoking, heavy alcohol consumption, and betel quid (BQ) chewing have been identi®ed as possible causes for the increased incidence [2,3]. Oral premalignant lesions (such as leukoplakia and erypla- kia) and malignant lesions (such as squamous cell car- cinoma) are frequently found among BQ chewers, giving urgency to the need to develop a screening and diagnostic technique for early-stage oral cancer. Among the diagnostic techniques for identifying oral premalignant and malignant lesions, pathological exa- mination of incisional biopsy specimen is the most reliable. Even experienced clinicians, however, cannot easily select a representative site for biopsy. Therefore, a technique for non-invasively detecting cancer or helping the clinician choose the appropriate site for biopsy can save patients from multiple biopsies and allow a broader range of diagnoses, which can aid early detec- tion of oral cancer. In contrast to conventional biopsy techniques, light- induced auto¯uorescence spectroscopy can be con- ducted to characterize tissues or to detect cancers with- out removing them, and the analysis results can be obtained in near real-time. Studies have shown that auto¯uorescence spectroscopy is useful for distinguish- ing cancerous from normal tissues in a variety of organ systems, including the human lung and breast [4,5], bronchus [6], colon [7,8], cervix [9,10], esophagus [11] and head and neck [12,13]. Some typical ¯uorophores in human tissues, such as proteins and amino acids (including tryptophan, tyrosin, phenylalanine, and elas- tin), purines, pyrimidines, and nucleic acids (including adenine, adenosine, guanine, guanosine, DNA, and RNA) absorb ultraviolet light near 280 nm and have auto¯uorescence peaks near 340, 360, and 390 nm [14,15]. This ®nding suggests that shorter excitation wavelengths, such as 280, 290, and 300 nm, could be ORAL ONCOLOGY Oral Oncology 35 (1999) 144±150 1368-8375/99/$Ðsee front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S1368-8375(98)00107-9 * Corresponding author. Tel.: +886-2-28267027; fax: +886-2- 28210847; e-mail: chiang@bme.ym.edu.tw