Classification of normal and cancerous lung tissues by electrical impendence tomography.

Abstract

Biological tissue impedance spectroscopy can provide rich physiological and pathological information by measuring the variation of the complex impedance of biological tissues under various frequencies of driven current. Electrical Impedance Tomography (EIT) technique can measure the impedance spectroscopy of biological tissue in medical field. Before application, a key problem must be solved on how to generally distinguish normal tissues from the cancerous in terms of measurable EIT data. In this paper, the impedance spectroscopy characteristics of human lung tissue are studied. On the basis of the measured data of 109 lung cancer patients, Cole-Cole Circle radius (CCCR) and the complex modulus are extracted. In terms of the two characteristics, 71.6% and 66.4% samples of cancerous and normal tissues can be correctly classified, respectively. Furthermore, two characteristics of the measured EIT data of each patient consist of a two-dimensional vector and all such vectors comprise a set of vectors. When classifying the vector set, the rate of correctly partitioning normal and cancerous tissues can be raised to 78.2%. The main factors to affect the classification results on normal and cancerous tissues are generally analyzed. The proposed method will play an important role in further working out an efficient and feasible diagnostic method for potential lung cancer patients, and provide theoretical basis and reference data for electrical impedance tomography technology in monitoring pulmonary function.

• Authors: Jianling Gao, Shihong Yue, Jun Chen, Huaxiang Wang
• Journal: Bio-medical materials and engineering (Biomed Mater Eng) Vol. 24 Issue 6 Pg. 2229-41 ( 2014) ISSN: 1878-3619 [Electronic] Netherlands
• PMID: 25226922 (Publication Type: Journal Article, Research Support, Non-U.S. Gov't)

Topics

• Adult
• Aged
• Aged, 80 and over
• Algorithms
• Diagnosis, Computer-Assisted (methods)
• Diagnosis, Differential
• Dielectric Spectroscopy (methods)
• Female
• Humans
• Lung Neoplasms (diagnosis, physiopathology)
• Male
• Middle Aged
• Pattern Recognition, Automated (methods)
• Reproducibility of Results
• Sensitivity and Specificity
• Tomography (methods)