The analysis of exfoliated cervical cells by infrared microscopy

The screening technique developed by Dr. Papanicolau for cervical cancer has significantly increased opportunities for early detection and treatment of this disease. Recently there has been a great deal of concern related to the accuracy of the Pap smear screening technique for cervical cancer and the need for highly skilled technologists to reliably examine the prepared samples. The use of infrared spectroscopy as a technique to discriminate between normal and malignant cervical cell samples has been reported by several research groups. Samples of cervical cells can be prepared for spectroscopic diagnosis by centrifuging cells obtained by the normal Pap smear technique and applying them to an infrared transparent window. A major concern with diagnosis by infrared spectroscopy is the homogeneity of the sample and whether the spectral features used in the analysis are caused by localized groups of cells or can be attributed to the majority of the cells in the sample. In this paper we report on the use of automated infrared microscopic mapping techniques to measure the infrared spectra at fixed intervals across the sample covering a 5 mm spot. Various post processing techniques have been applied to the spectral results to create images revealing specific features of the sample. The techniques applied to the data include: 1) Baseline corrected peak height; 2) Band ratioing to compensate for thickness; 3) Correlation metrics; 4) Similarity matching, and 5) Mahalanobis distance classification algorithm.