Summary of the impact

This case study describes how computational research in boundary problems at Middlesex was applied to bio-imaging using Electrical Impedance Tomography (EIT) for imaging brain function, lung function and tumour detection, and the development of Optical Tomography of brain function in neonates. This has resulted in the contribution of several public domain, open source resources to the international industrial and commercial research communities such as novel reconstruction algorithms, geometric models for generating accurate finite element and boundary element forward models and methods to generate subject-specific forward models which have been exploited as detailed below. It has also resulted in two patents.

Underpinning research

The development of bio-imaging using Electrical Impedance Tomography (EIT), is a low-cost, portable, non-invasive, non-radiating technique allowing the subject to move during scanning. It is therefore ideal for clinical applications and a multidisciplinary group at a Middlesex comprising computer scientists, health professionals and medical physicists has been developing this method for over 20 years. Measured conductivity can be presented as either a 2D or 3D image for clinical evaluation and diagnosis of a range of conditions associated with gastric emptying, breast tumour localisation, lung function and brain function. EIT poses an inverse problem, and the group has focussed its computational work on demonstrating that a valid and viable forward model of target geometry and conducting priors is required for successful implementation, developing methods that present increasingly accurate forward models, developing associated algorithms and exploring potential clinical application.

In collaboration with the Medical Physics group at University College, a significant outcome of this work was the generation of the first successful 2D images of impedance change inside the human head [1]. The main motivation of the work was to develop EIT for localising epileptic foci for possible surgical intervention. Since 1999, Dr Andrew Tizzard has been working on improving the forward models for the human head, one outcome being the generation of the underlying geometry and finite element (FE) models of adult and neonate heads enabling the progress of the algorithm development to imaging in 3D. Tizzard and Bayford also developed a method and associated algorithms for automatically generating subject-specific FE models through elastic deformation from electrode position data [2].

This led to the key finding that boundary form is a major factor in reducing computational artefacts in the reconstructed image. The research produced the first image of epileptic focus in a human using EIT [3]. Bayford and Tizzard have developed this research further for monitoring neonate lung function using a wavelet AMG and estimated boundary form [4]. They also contributed significantly to an international collaboration for a consensus algorithm for EIT of lung function and extended the research to image rectal cancer. The research on boundary form has provided the key for improving this imaging technique.

Paper [3] below was an extensive international collaboration instigated by Bayford, Tizzard and colleagues from a number of research groups worldwide and features a significant contribution by Tizzard for the accurate forward model of the adult thorax. It describes a consensus algorithm intended to be a benchmark for comparing future developments of EIT algorithms for monitoring lung function. The framework for the algorithm consists of: 1) detailed finite element models of a representative adult and neonatal thorax; 2) consensus on the performance figures of merit for EIT image reconstruction; and 3) a systematic approach to optimize a linear reconstruction matrix to desired performance measures.

Paper [6] summarises much of this work at Middlesex and resulted in the software contributed to the EIDORS project (eidors3d.sourceforge.net/).

References to the research

The research was based on major, competitively funded peer reviewed research, led to two patents, and was published in peer reviewed journals of high repute as detailed below:

Grants:

2009 EPSRC EP/G061572/1 New imaging methods for the detection of cancer biomarkers (£800K) Total grant for all partners £1.7M

2007 EPSRC EP/E031633/1, A Novel Analogue Bio-impedance system-on-a-chip for monitoring of neonate lung function (£182466) Total for all partners 600K

Patents:

(WO/2010/052503) Detection of Cancer

(61/867,904 : US) "A flexible wearable device for the acquisition of data for EIT of lung function"

Details of the impact

Work by Bayford and Tizzard on the development of algorithms and the generation of geometrical finite element forward models, applied to imaging impedance changes inside the human brain, has led to impact in three areas. Firstly, the availability of the software has led to its adoption in commercial software. Secondly, work on neonate lung function monitoring has generated a patent. Thirdly, the software has been used by both industrial and academic research groups for the provision of further imaging solutions such as in improving breast tumour images. Some of this work is commercially confidential.

(i) The algorithm has been made available via the EIDORS system (eidors3d.sourceforge.net/) and has over 1000 downloads [S1, S2, S7]. It is now used extensively by clinicians and researchers in industry and research hospitals globally. The use of the algorithm has led to a significant improvement to the only existing commercial medical EIT system originally developed by Viasys. (now Carefusion). Dräger, a leading international company in the fields of medical and safety technology (http://campaigns.draeger.com/pulmovista500/en/) has also adopted the EIDORS software [S6].

(ii) Work on neonate lung function monitoring by Middlesex and collaborators at UCL,City University and GOSH has led to a patent being filed in the US (61/867,904) that describes a flexible wearable device to extract boundary information for the warping algorithm to dynamically generate and modify subject specific forward models in real time [S3]. This work addresses the urgent need for objective, non-invasive measures of lung maturity and development, oxygen requirements and lung function, suitable for use in small, un-sedated infants, to define the nature and severity of lung dysfunction. We have signed an NDA with Swisstom to develop the wearable device [note: this is commercially confidential information] and have a provisional patent in place (61/867,904: US) [S3].

We are also working with Midatech Ltd to extend the application of EIT for the detection of cancer. A joint patent (WO/2010/052503) Detection of Cancer) is in place with this company, which is led by Prof. Rademacher [S5]. This work is also subject to an NDA, which limits the information we are allowed to disclose in this document.

(iii) The models generated have also been used extensively internationally by other clinical development groups focused on the production of imaging solutions. For example clinicians at Dartmouth College and Florida University in the USA are developing EIT and optical tomography imaging of both adult and neonate human heads, making extensive use of our algorithms downloaded from EIDORS (eidors3d.sourceforge.net/). The automatic generation of subject- specific forward models, the warping algorithm, has formed the basis of extending the area of impact to imaging lung function, specifically in neonates, in collaboration with Prof. Andreas Demosthenous, Dept. Electronic and Electrical Engineering, UCL, Prof. Panos Liatsis, City University and Prof. Janet Stocks, Great Ormond St. Hospital. The work has also been used to improve breast tumour imaging in collaboration with Prof. Alex Hartov's team that includes Dr Andrea Borsic and Prof. Ryan Halter at Dartmouth College. This work is based on using elastic deformation to warp standard or idealised geometry — all of which provides extensions of the public domain toolset [S7].

Use of the research outcomes for imaging neonate lung function significantly reduces the cost of patient care and mortality by allowing systematic monitoring of child lung function for time-critical intervention. There is an urgent need to improve ventilation strategies in children with acute lung injury (ALI), which has a high mortality (22%) compared with the overall mortality of paediatric intensive care unit patients. Disorders of lung growth, maturation and control of breathing are among the most important problems faced by the neonatologist. Although it is not yet in mainstream clinical use for monitoring lung function in neonates, our algorithms are being used for adult patients [S6] using some of the developments created for neonate imaging.

Sources to corroborate the impact

S1. Andy Adler and William R.B. Lionheart (2006) Uses and abuses of EIDORS: An extensible software base for EIT. Physiological Measurement, 27. S25-S42 [Confirmation on the use of the EIDORS software]

S2. Prof. Andy Adler P.Eng. [manager of the EIDORS software at Sourceforge]. Systems and Computer Engineering, Carleton University, Ottawa, Canada. Prof. Adler manages the EIDORS software repository, and will be able to confirm that we have contributed to it and that it can be used by industrial users.

S3. Paul Casbon, Lucas & Co., Chartered Patent Attorney; European Patent Attorney (re. 61/867,904: US). Can corroborate the filing of the provisional US patent.

S4. Prof. Alex Hartov, Professor of Engineering, Dartmouth College, Hanover, USA. Prof Hartov can confirm that we have been working on boundary shape for breast imaging with them and that we have modified the work on the lung to accommodate this.

S5. Prof. Rademacher (Chief Scientific Officer and Chairman), Midatech Ltd., Oxfordshire OX13 6BH, UK. Prof. Rademacher can confirm that we hold a joint patent in EIT and are working on the clinical application of EIT within the limits of the NDA agreement we have with him.

S6. Dräger (Drägerwerk AG & Co.): "Regional lung function monitoring — how EIT can help to guide ventilation therapy"; http://campaigns.draeger.com/pulmovista500/en/

S7. EIDORS: Electrical Impedance Tomography and Diffuse Optical Tomography Reconstruction Software; http://eidors3d.sourceforge.net/