Quantitative chromatics analysis for computer imaging of cytologic subtypes of lung cancer stained by Papanicolaou stain

OBJECTIVE: To determine the role of quantitative chromatics analysis in the classification of subtypes of lung cancer stained by Papanicolaou stain. STUDY DESIGN: By means of computer image analysis, 60 keratinized squamous carcinoma cells (KSCC), 88 nonkeratinized squamous carcinoma cells (NKSCC) and 150 adenocarcinoma cells (ACC) from lung cancer in sputum smears stained by Papanicolaou stain were analyzed and distinguished based on quantitative colorimetry. The features measured were the content of three primary colors, red (R), green (G) and blue (B) and the coefficients of R, G and B (r, g and b, respectively). Hue, saturation, brightness and gray level were also measured. A stepwise discriminant analysis was carried out. RESULTS: The values of R, G and B and r, g and b, hue and saturation in NKSCC and ACC were significantly different from those of KSCC, and the changes in the three primary colors were more sensitive than those in the gray level. Computer assessment based on three primary color coefficients, hue and saturation yielded accuracy of distinguishing KSCC from NKSCC and KSCC from ACC of 95.2% and 95%, respectively. CONCLUSION: Quantitative analyses of R, G and B and r, g, b and hue and saturation are valuable in distinguishing KSCC from NKSCC and ACC.