Advanced digital electrical impedance tomography system for biomedical imaging

Research output: ThesisDoctoral Thesis

Abstract / Description of output

Electrical Impedance Tomography (EIT) images the spatial conductivity distribution in an electrode-bounded sensing domain by non-intrusively generating an electric field and measuring the induced boundary voltage. Since its emergence, it
has attracted ample interest in the field of biomedical imaging owing to its fast, cost efficient, label-free and non-intrusive sensing ability. Well-investigated biomedical applications of the EIT include lung ventilation monitoring, breast cancer imaging, and brain function imaging. This thesis probes an emerging biomedical application of EIT in three dimensional (3D) cell culture imaging to study non-destructively the biological behaviour of a 3D cell culture system, on which occasion real-time qualitative and quantitative imaging are becoming increasingly desirable. Focused on this topic, the contribution of the thesis can be summarised from the perspectives of biomedical-designed EIT system, fast and effective image reconstruction algorithms, miniature EIT sensors and experimental studies on cell imaging and cell-drug response monitoring, as follows. First of all, in order to facilitate fast, broadband and real-time 3D conductivity imaging for biomedical applications, the design and evaluation of a novel multi-frequency EIT (mfEIT) system was presented. The system integrated 32 electrode interfaces and its working frequency ranged from 10 kHz to 1 MHz. Novel features of the system included: a) a fully adjustable multi-frequency current source with current monitoring function was designed; b) a flexible switching scheme together with a semi-parallel data acquisition architecture was developed for high-frame-rate data acquisition; c) multi-frequency simultaneous digital quadrature demodulation was accomplished, and d) a 3D imaging software, i.e. Visual Tomography, was developed to perform real-time two dimensional (2D) and 3D image reconstruction, visualisation and analysis. The mfEIT system was systematically tested and evaluated on the basis of the Signal to Noise Ratio (SNR), frame rate, and 2D and 3D multi-frequency phantom imaging. The highest SNR achieved by the system was 82.82 dB on a 16-electrode EIT sensor. The frame rate was up to 546 frames per second (fps) at serial mode and 1014 fps at semi-parallel mode. The evaluation results indicate that the presented mfEIT system is a powerful
tool for real-time 2D and 3D biomedical imaging.
The quality of tomographic images is of great significance for performing
qualitative or quantitative analysis in biomedical applications. To realise high quality
conductivity imaging, two novel image reconstruction algorithms using adaptive
group sparsity constraint were proposed. The proposed algorithms considered both
the underlying structure of the conductivity distribution and sparsity priors in order
to reduce the degree of freedom and pursue solutions with the group sparsity
structure. The global characteristic of inclusion boundaries was studied as well by
imposing the total variation constraint on the whole image. In addition, two adaptive
pixel grouping methods were also presented to extract the structure information
without requiring any a priori knowledge. The proposed algorithms were evaluated
comparatively through numerical simulation and phantom experiments. Compared
with the state-of-the-art algorithms such as l1 regularisation, the proposed algorithms
demonstrated superior spatial resolution and preferable noise reduction performance
in the reconstructed images. These features were demanded urgently in biomedical
imaging.
Further, a planar miniature EIT sensor amenable to the standard 3D cell
culture format was designed and a 3D forward model was developed for 3D imaging.
A novel 3D-Laplacian and sparsity joint regularisation algorithm was proposed for
enhanced 3D image reconstruction. Simulated phantoms with spheres located at
different vertical and horizontal positions were imaged for 3D imaging performance
evaluation. Image reconstructions of MCF-7 human breast cancer cell spheroids and
triangular breast cancer cell pellets were carried out for experimental verification.
The results confirmed that robust impedance measurement on the highly conductive
cell culture medium was feasible and, greatly improved image quality was obtained
by using the proposed regularisation method.
Finally, a series of cancer cell spheroid imaging tests and real-time cell-drug
response monitoring experiments by using the developed mfEIT system (Chapter 3),
the designed miniature EIT sensors (Chapter 6) and the proposed image
reconstruction algorithms (Chapter 4, 5 and 6) were carried out followed by
comparative analysis. The stability of long-term impedance measurement on the
highly conductive cell culture medium was verified firstly. Subsequently, by using
the proposed algorithms in Chapter 4 and Chapter 5, high quality cancer cell
spheroid imaging on a miniature sensor with 2D electrode configuration was
achieved. Further, preliminary experiments on real-time monitoring of human breast
cancer cell and anti-cancer drug response were performed and analysed. Promising
results were obtained from these experiments.
In summary, the work demonstrated in this thesis validated the feasibility of
using the developed mfEIT system, the proposed image reconstruction algorithms, as
well as the designed miniature EIT sensors to visualise 3D cell culture systems such
as cell spheroids or artificial tissues and organs. The established work would
expedite the real-time qualitative and quantitative imaging of 3D cell culture systems
for the rapid assessment of cellular dynamics.
Original languageEnglish
QualificationPh.D.
Awarding Institution
  • University of Edinburgh
Supervisors/Advisors
  • Jia, Jiabin, Supervisor
  • Polydorides, Nick, Supervisor
Award date4 Jul 2018
Publication statusPublished - Jul 2018

Fingerprint

Dive into the research topics of 'Advanced digital electrical impedance tomography system for biomedical imaging'. Together they form a unique fingerprint.

Cite this