Related Experiment Video
Updated: Jul 14, 2026
15:48
Tracking the Mammary Architectural Features and Detecting Breast Cancer with Magnetic Resonance Diffusion Tensor Imaging
Published on: December 15, 2014
Classification of breast tumour using electrical impedance and machine learning techniques
Abdullah Al Amin1, Shahnaj Parvin, M A Kadir
1Department of Biomedical Physics and Technology, University of Dhaka, Dhaka 1000, Bangladesh.
Physiological Measurement
|May 22, 2014
Summary
This study explored using non-invasive electrical impedance measurements and machine learning to differentiate malignant and benign breast tumors. Preliminary results show promising accuracy, potentially offering a less invasive alternative to biopsy.
Area of Science:
- Biomedical Engineering
- Medical Physics
- Oncology
Background:
- Breast lump characterization typically relies on invasive biopsy.
- Current diagnostic methods carry surgical risks.
- Non-invasive techniques are sought to improve breast cancer diagnosis.
Purpose of the Study:
- To investigate the feasibility of using electrical impedance measurements and machine learning for non-invasive tumor characterization.
- To assess the potential of this technique as an alternative to traditional biopsy.
Main Methods:
- Utilized tetrapolar impedance measurement (TPIM) at 5 and 200 kHz in two orthogonal directions.
- Applied machine learning (K-NN classification) to impedance data from 19 subjects.
- Extracted 12 features, including frequency-dependent impedance and age-adjusted values.
Main Results:
- Feature plots showed overlap between malignant and benign tumors initially.
- Age-adjusted impedance features improved tumor separation.
- K-NN classification achieved high negative prediction value (93%) and specificity (87%).
Conclusions:
- Electrical impedance measurements combined with machine learning show potential for non-invasive breast tumor characterization.
- The technique demonstrated good efficacy (84%) in a small sample size.
- Further studies with larger cohorts are needed to validate and refine this promising diagnostic approach.
