The Cole relaxation frequency as a parameter to identify cancer in breast tissue.

PURPOSE: To correlate the Cole relaxation frequencies obtained from measurements of the electrical properties of breast tissue to the presence or absence of cancer. METHODS: Four-lead impedance measurements were obtained on ex vivo specimens extracted during surgery from 187 volunteer patients. Data were acquired with a commercial Solartron impedance bridge employing 4-lead Ag-AgCl or blackened platinum (BPt) electrodes at frequencies logarithmically spaced from 1 Hz to 3.2 × 10(7) Hz utilizing 6-10 frequencies per decade. The Cole frequencies obtained from these measurements were correlated with the tissue health status (cancer or noncancer) obtained from histological analysis of the specimens. RESULTS: Analysis of the impedance measurements showed that the Cole relaxation frequencies correlated to the presence or absence of cancer in the examined tissue with a sensitivity up to 100% (95% CI, 99%-100%) and a specificity up to 85% (95% CI, 79%-91%) based on the ROC curve of the data with the Cole frequency as the classifier. CONCLUSIONS: The results show that the Cole frequency alone is a viable classifier for malignant breast anomalies. Results of the current work are consistent with recent bioimpedance measurements on single cell and cell suspension breast cell lines.

Correction of electrode polarization contributions to the dielectric properties of normal and cancerous breast tissues at audio/radiofrequencies.

Spurious contributions from electrode polarization (EP) are a major nuisance in dielectric measurements of biological tissues and hamper accurate determination of tissue properties in the audio/radiofrequencies. Various electrode geometries and/or treatments have been employed traditionally to reduce EP contributions, although none succeeded to completely remove EP from measurements on tissues for all practical frequency ranges. A method of correction for contributions of EP to the dielectric properties of tissues is proposed. The method is based on modeling the electrode impedance with suitable functions and on the observation that certain parameters are only dependent on electrodes properties and can thus be determined separately. The method is tested on various samples with known properties, and its usefulness is demonstrated with samples of normal and cancerous human female breast tissue. It is observed that the dielectric properties of the tissues over the frequency range 40 Hz-100 MHz are significantly different among different types of breast tissue. This observation is used further to demonstrate that, by scanning the tip of the measuring dielectric probe (with modest spatial resolution) across a sample of excised breast tissue, significant variations in the electrical properties are detected at a position where a tumor is located. This study shows that dielectric spectroscopy has the potential to offer a viable alternative to the current methods for detection of breast cancer in vivo.