A clinical study of the use of impedance spectroscopy in the detection of cervical intraepithelial neoplasia (CIN)

Abstract

A clinical study of the use of impedance spectroscopy in the detection of cervical intraepithelial neoplasia (CIN) S. Abdul, B.H. Brown, P. Milnes, J.A. Tidy a Sheffield Gynaecological Oncology Centre, Royal Hallamshire Hospital, Sheffield, UK Medical Physics, Royal Hallamshire Hospital, Sheffield, UK Fig. 2. Modelled impedance spectra for tissue types. Background. A 4-electrode impedance probe, 5.5 mm in diameter (Fig. 1), with 2 electrodes injecting a current of 20 AA and the other 2 electrodes measuring the impedance spectrum, at 8 frequencies has been shown be a promising cervical screening tool [1,2]. Good separation of CIN from squamous epithelium using this probe has been achieved, but separation from immature metaplastic tissue was less defined. Finite element modelling [3] suggests that improved separation of CIN from normal tissue may be achieved using a wider frequency range (Fig. 2). The mark III impedance probe has been designed to take 30 measurements over a frequency range of 2–1200 kHz. Objective. To assess the performance of the new cervical impedance spectroscopy mark III probe in the separation of CIN from normal tissue. 0090-8258/$ see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ygyno.2005.07.046 i Abstract submitted for the Fourth International Cervical Cancer Conference. Fig. 1. Impedance probe. Design. Prospective observational study of 176 women referred for colposcopy with an abnormal smear (borderline, mild, moderate or severe dyskaryosis). 99 (2005) S64 – S66 Fig. 3. Measurement points on cervix. YGYNO-71096; No of Pages 3; 4C: 1, 2 Fig. 4. Measured impedance spectra for tissue types. Abstract S65 Method. Electrical impedance spectra were recorded from 8 points on the cervix (Fig. 3). Colposcopy examination and probe positioning were video recorded for analysis to allow classification of each point into different epithelia types. Impedance spectra measured (Fig. 4) were fitted by a method of least square regression to Coles’ equation to give parameter R, S, C, Fc and R/S for the different epithelia. Outcome measures. Cervical impedance derived parameters R, S, C, Fc and R/S in different epithelia and the performance of the probe in identifying women with CIN. Results. 176 women were assessed. 1360 measurements were classified as squamous epithelium (680), CIN 1 (39), CIN 2/3(178), mature metaplasia (135), immature metaplasia (79) and columnar epithelium (28). From normal epithelium through CIN 1 to CIN 2/3, R decreases significantly (P < 0.001 Mann Whitney test) by a factor of 3.2, S increased significantly (P < 0.0001 Mann Whitney test) by a factor of 2.0, Fc increases by a factor of 4.5, R/S decreases by a factor of 4.0 and C does not change (Fig. 5). Mann Whitney tests showed several significant separations (P < 0.001). CIN 2/3 can be separated from: normal Fig. 5. R, S and Fc values in tissue types. squamous epithelium (by R, S, C, Fc and R/S); columnar epithelium (by R, S, C, Fc and R/S); and mature metaplastic tissue (by R, S, C, Fc and R/S). Separation of CIN 2/3 from immature metaplasia could not be achieved by any parameter. CIN 1 can be separated from: normal squamous epithelium (by R, S, Fc and R/S); columnar and immature metaplastic tissue (by R, S, C, Fc and R/S); and CIN 2/3 (by R, S, C, Fc and R/S). Receiver operating characteristics (ROC) curves were derived for separating CIN from normal epithelium (Fig. 6). The area under the curve (AUC) values are given for separating CIN 2/3 (using R, S, C, Fc and R/S respectively) from: original squamous epithelium (0.88, 0.83, 0.63, 0.79 and 0.89 respectively); columnar epithelium (0.68, 0.65, 0.67, 0.67 and 0.69 respectively); mature metaplastic epithelium (0.80, 0.80, 0.69, 0.77 and 0.81 respectively); Fig. 6. ROC curve using R to separate squamous epithelium from CIN2/3 (Area Under Curve = Curve 0.88).