The primary reason for conducting this research is to lay the foundations for developing a new non-invasive, atraumatic and painless system for breast tumour diagnosis and location. The proposed technique combines earlier concepts of ultrasonic detection of elastic variations with newly available tissue Doppler ultrasound systems. It includes the concept of detecting time-dependent elasticity (viscoelasticity) rather than static elasticity, which may provide better contrast.
The commonest clinical method for detecting lumps within tissue, palpation, is highly subjective and dependent on the skill of the practitioner. The method exists because certain pathological conditions, such as malignant tumours, manifest themselves as changes in the tissue’s mechanical stiffness. While X-ray imaging is well established for the detection of small, deeply located tumours, X-ray hazards and the desire for better performance have led to a continuing search for alternative techniques. Diagnostic ultrasound is a potential alternative to X-rays, its limitation being that small pathological changes in tissue are difficult to discern on normal ultrasound B-scans. If, however, ultrasonic echo data is collected before and after a slight compression of the tissue (elastography) comparisons can be made between normal and pathological areas. This is possible because normal tissues will exhibit relatively more movement than stiffer pathological regions. It has been suggested [1] that benign and malignant tumours can be distinguished by elastography due to their differing uniformity of elastic properties.
Elastography examines the static elastic properties of tissue. Other authors and unpublished results from our own tissue property studies suggest that differences between healthy and pathological tissue are highlighted more clearly using rapidly changing strain. This time-dependent (i.e. viscous) response is analogous to the vibrational frequency response of the tissue. It is likely that a relatively narrow band of vibration frequencies exists for which the response in the tissue is optimum for distinguishing variations in viscoelastic properties. Imaging of elastic properties in the presence of vibration is known as sonoelastography.
Several authors have reported the use of phase-contrast magnetic resonance imaging (MRI) to visualise the mechanical properties of tissues [2,3]. Here, images of tissue subjected to static or time varying displacement are obtained, yielding information on the 3D distribution of elasticity and viscoelasticity respectively. The results from these techniques are very impressive in that small inhomogeneities can be localised. However, the method may never become broadly applicable due to the very high cost of MRI and its lack of portability.
In Doppler ultrasonography, frequency shifts in returning ultrasonic echoes due to the motion of reflecting features are detected. While this is commonly used to highlight blood flow, variations in tissue motion can also be revealed. If a vibration (10-1000Hz) is applied to the tissue under inspection, a small stiff zone will appear as a defined region due to the difference between its motion and that of the surrounding tissue.
Ultrasound sonoelastography imaging has been compared to conventional ultrasound imaging [4] for the detection of prostate cancer in vitro, with promising results. Although elastography and sonoelastography are not yet being used in routine clinical practice, these imaging methods have the potential to give comparable spatial resolution to standard grey-scale imaging with enhanced tissue discrimination.
Standard 2D ultrasound scanners can provide 3D images if 6 degrees-of-freedom (6DOF) measurements of probe position and orientation are recorded simultaneously. This was demonstrated for a free-hand technique in which the physician performs an examination in a normal manner [5,6]. Existing visualisation software can provide an operator with the ability to explore the 3D images using techniques such as any-slice imaging, rendering with transparency and 3D viewpoint control. Probe position can be sensed using a Polhemus Fastrak electromagnetic sensor, with an angular and positional accuracy of 0.5 deg. and 0.5mm respectively.
Its effectiveness in freehand scanning has been demonstrated [5,7]. The position information is used to assemble the B-scan images into a 3D image composed of a regular voxel array. In practice, errors are introduced by inaccurate 6DOF measurements and to reduce these errors the 2D images need to be aligned, using a registration algorithm, prior to being combined [7]. Interpolation is used to fill the 3D volume while data is fused where B-scans intersect. This fusion process, known as ‘compounding’, has the advantage of reducing speckle noise and shadowing, and leads to improved image quality [8].
Work combining sonoelastography with 3D ultrasound image reconstruction by is being conducted by very few researchers worldwide, limited mainly to those quoted above. Our proposal to apply 3D reconstruction to Doppler ultrasound images is new. While reconstruction with standard ultrasound scans has been successfully demonstrated, use of Doppler ultrasound data brings added problems of signal loss and noise presence. We will attempt to solve this problem by averaging over larger quantities of data and by improving the processing algorithms. To our knowledge the group, based in Rochester NY [4], is the only centre investigating sonoelastography using a combination of tissue vibration and ultrasound Doppler imaging. This group has demonstrated the feasibility of the technique despite the difficulty of delivering tissue vibration to deep regions (e.g. prostate). In contrast to the Rochester work, we intend to focus on breast lesions, where it should be easier to deliver the vibrations and directly scan the tissue, and to use 3D image reconstruction of the Doppler ultrasound data.
The medical relevance of the proposal is twofold. Firstly, a positive outcome of the project could lead to a new imaging modality that is painless and risk free in use with equipment that is relatively cheap and portable. Secondly, advances that we may make in 3D ultrasound imaging would be widely applicable to general ultrasound imaging. An additional potential application, deriving from the work, is measurement of the volume distribution of tissue elastic properties. This data is required by researchers developing mathematical models of tissue and, in particular, for using in electronic tissue representation in surgical simulators.
We will investigate methods for exciting vibration in tissue over a range of frequencies and amplitudes. Simultaneously, we will develop the software and hardware aspects of the 3D reconstruction system. Towards the end of the first year we will conduct studies to determine the optimum frequencies and amplitudes of vibration required to distinguish areas of differing elasticity within (i) tissue phantoms and (ii) mastectomy specimens. The project will be completed by a period of evaluation and system refinement. Details of this programme are given below. Workpackages, with measurable objectives, are supplied below and a Gantt chart provides the project timetable and employee resource allocation.
Free-hand scanning and vibration system: There are two systems, a Polhemus electro-magnetic device and a Kinemetrix infra-red motion tracking system, available to the project for acquiring 6DOF probe position and orientation data. Each has potential advantages and both will be investigated. The Polhemus device would lead to a highly portable system although it would be sensitive to electro-magnetic interference. The IR tracking system is less portable but does not suffer from interference. In either case, the device must be calibrated with respect to the B-scans using an object with known geometry. We will develop and evaluate a vibration system, consisting of a signal source with variable frequency (10-1000Hz) and amplitude, amplifier and transducer. The transducer development will focus on comparing loudspeaker devices with other vibration sources. In particular, the coupling (e.g. air, air bellows, pliable foam etc.) between the transducer and tissue will be investigated. Solving this mechanical impedance matching problem will overlap with studies to find optimal frequency ranges for detecting regions of altered tissue stiffness.
Incremental Development: We will develop and evaluate the system incrementally starting with the use of commercially available synthetic tissue phantoms (Siel Imaging Ltd, Aldermaston, Berks.) before progressing to mastectomy specimens. By inserting structures of known geometry and position in the phantoms, we will check the accuracy of the 3-D reconstruction software both for normal ultrasound scans and for tissue Doppler scans.
Reconstruction and Visualisation: An Aloka SSD2200 ultrasound scanner with a tissue Doppler system will be used. (The scanner is available to the project but the Doppler probe will need to be purchased.) We aim to achieve high quality 3D reconstruction of sonoelastographic data through the use of (i) accurate calibration of the free-hand scanning device, (ii) accurate registration, and (iii) spatial interpolation and data fusion. We will fuse multiple 3D data sets, scanned with different vibration source positions, to improve the quality of the reconstruction by reducing speckle noise and other artefacts such as shadowing. Techniques such as alignment by maximisation of mutual information [9] will be investigated for image registration. Publicly available software ("StradX") could be used for initial prototype development [7,8]. Current 3D ultrasound techniques will be further developed to cope with the significant signal dropout and noise present in Doppler ultrasound data. Sonoelastographic scanning requires the determination of optimal frequency ranges for the externally applied vibration. We propose to partially automate a frequency search based on statistical analysis of the resulting Doppler scans. We will implement visualisation of voxel uncertainty based on the amount of evidence available from the 2D scans and the extent of the interpolation used. This will provide useful information for subsequent data fusion and provide feedback to the physician performing the scan. Evaluation criteria will be (i) absolute accuracy assessed using phantoms of known geometry, (ii) consistency and repeatability, and (iii) ease of use.
Workpackages and Objectives
WP1 Tracking and data management (RA1, 12 weeks)
1. Investigate the Polhemus and Kinemetrix systems for measurement of hand-piece position and orientation.
2. Set-up the acquisition system including calibration and synchronisation with the ultrasound scanner.
3. Capture multiple sets of 2D and Doppler ultrasound scans along with synchronised 6DOF data.
4. Convert and display image sets on development workstation.
WP2 Reconstruction of 3D Doppler ultrasound data (RA1, 40 weeks)
1. Demonstrate 3D reconstruction from 2D scans based on 6DOF data.
2. Implement and improve image registration and spatial compounding to optimise 3D image quality
3. Develop image visualisation (any-plane slicing, volume rendering with variable transparency, animation).
WP3 Vibration system development (RA2, 40 weeks)
1. Develop the source and transmission system for delivering vibration to tissue.
WP4 Phantom based evaluation of prototype system (RA2, 16 weeks)
1. Determine the smallest detectable lesion size and evaluate frequency dependence in phantoms on 2D scans.
2. Develop semi-automatic frequency selection based on statistical analysis of Doppler ultrasound images.
3. Evaluate accuracy of 3D reconstruction based on the known phantom geometry (at end of workpackage).
WP5 Human tissue (mastectomy specimens) based evaluation (RA1 & RA2, 16 weeks)
1. Determine optimal frequencies for detection of tumours in mastectomy specimens.
2. Perform 3D reconstruction and visualisation of specimens to determine position and size of tumours.
WP6 Final evaluation and report (RA1 & RA2, 15 weeks)
1. Evaluate the final system and produce a report containing the final specification and recommendations.
References
The Department of Molecular Oncology headed by Prof. D Lane has amalgamated with the Department of Surgery to form the new Department of Surgery and Molecular Oncology with Prof. Sir Alfred Cuschieri as the Head of the new Department. These two Departments, their research Groups and Principal Investigators have an established track record and international reputation for research in Molecular Oncology, Minimal Access Surgery and Clinical Oncology. Surgery has been ‘star’ rated in the UK Research Assessment Exercise since the inception of this ranking system. There are five Sections in the New Department: Surgical Skills Unit, The Surgical Technology Group, the MAS group, Molecular Oncology ( Prof D Lane) and Surgical Oncology (Prof R Steele).
The Surgical Skills Unit (SSU) runs major clinical, research and training programmes in MAS and has just received ‘The Queen’s award’ for Higher Education in 1999. The SSU is the national training unit for MAS in Scotland and has a full-time faculty of lecturers and technician tutors. The unit also undertakes research in the development of simulation techniques for skills training and in technology for assessment of psychomotor skills for surgery.
The Surgical Technology Group headed by Dr T Frank and Sir Alfred Cuschieri is devoted to R & D in MAS and has been funded by MRC, Wellcome and other Charities, Defence Advanced Research Products Agency (DARPA - USA), DTI, MedLINK, Scottish Office and Industry. Its activities include design and development of complex intelligent endomanipulators, visual display technology, shape memory constructs (sutures and ligatures) and their applicators for endoscopic surgery, gas-less laparoscopy techniques, biomechanical properties of benign and malignant tissues etc. At present, there are two full-time members of staff in the Department dedicated to research in biomechanical studies of soft tissues. The work has covered direct indentation measurements of the surface of solid organs and tension tests of hollow organs. These experimental data have provided the basis for the formulation of mathematical models that describe surface deformations of solid organs (work in collaboration with Dr P Davies and Dr P Connor, Mathematics Department, University of Strathclyde). More recently, we have developed instrumentation to gather tissue elasticity data from patients during open surgery in order to refine the mathematical models. The proposed work is a logical extension to us, as we intend to progress from single point compliance to direct measurement of the 3-dimensional distribution of tissue compliance variation.
The head of the Surgical Technology Group (Dr Frank) is a physicist with extensive experience in mechanical and electromechanical design (including an automatic medical percussion sound system using computer controlled impulse delivery and computational sound spectrum analysis). The group also includes a post-graduate engineer, Mr McLean (named in the proposal), who designed and constructed our tissue compliance probes. These devices are motorised, computer controlled and usable on human subjects during open or laparoscopic surgery. The group’s research is supported by ‘in-house’ design and manufacturing activities. Computer-aided design (CAD) is used during device development and our workshop has computer controlled milling and turning facilities linked to the CAD system.
The Clinical MAS group headed by Sir Alfred Cuschieri is concerned with phase II and Phase III studies in MAS.
Molecular Oncology headed by Prof David Lane is really a major Division rather than a Section with 40 research molecular biochemists who interact closely with the Surgical Oncologists for effective translational research on p53 based treatment of certain tumours. We are on track to start clinical trials in patients with metastatic liver disease from primary colorectal cancer towards the end of this year.
The Surgical Oncology Section headed by Prof. Steele has clinical and research interest in Gastrointestinal, hepatobiliary and breast cancer (four senior academic surgeons). The group has special research interest in metalloproteinases in cancer (Prof. Steele), molecular oncology (Mr A Thompson), in-situ ablation under laparoscopic/contact ultrasound guidance (Sir Alfred Cuschieri and Mr S Shimi).
The Department of Applied Computing at the University of Dundee was awarded a "5A" rating in the 1996 UK Research Assessment Exercise and thereby ranked in the top 3 computing departments in Scotland and in the top 16 in the UK. It comprises 15 academic staff and 15 research staff and currently has an annual research income £0.5M. Professor Ricketts holds a personal chair in Assistive Systems and Healthcare Computing within the Department of Applied Computing and Dr Stephen McKenna is a recently appointed lecturer with strong research record in computer vision. Professor Ricketts is currently co-holder of 6 research grants in medical informatics amounting to over £0.75M