Improvement of Breast Cancer Tactile Imaging through Non-Rigid Mosaicing

Lead Research Organisation: University of Strathclyde
Department Name: Electronic and Electrical Engineering

Abstract

This proposal considers non-rigid image registration and data fusion to localise a hand-held probe as it is scanned over the human body. To give the project context and provide a clear route to impact the project will focus on the specific case of tactile breast imaging, an exciting new methodology for breast cancer imaging. That said, the long term impact of this project will extend into other medical applications and general computer vision research.

The project will demonstrate the ability to localise a handheld probe scanned over the breast by imaging the vascular network and registering images using a finite element guided deformable registration model. The system will be demonstrated using a palpitation phantom and its performance fully characterised using metrology equipment available at Strathclyde (funded through EP/G038627/1).

The proposed target application, tactile imaging, creates a stress image that can be used to characterise breast cancer lesions, providing estimates of their size and relative elasticity. Advantages of tactile breast imaging over other imaging modalities include inherently low cost, ease-of-use, portability, and minimal training requirements. Unlike mammography which provides a complete image of the breast, tactile imaging probes are scanned over the breast, producing a real time feed of the pressure profile under the probe. This temporal feed is more difficult to interpret than a global image of the breast as the data has less spatial context. Furthermore, the scanning approach risks missing areas of the breast as it relies on the examiner to visit all regions with no feedback that this has been achieved. Representing the results as a global mosaic would significantly de-skill the procedure broadening the suitability of the technology.

Planned Impact

The proposed research will consider non-rigid image registration and data fusion to localise a hand-held probe as it is scanned over the human body. To give the project context and provide a clear route to impact the project will focus on the specific case of tactile breast imaging, an exciting new methodology for breast cancer imaging. That said, the long term impact of this project will extend into other medical applications and general computer vision research.

Breast cancer accounts for nearly one quarter of all cancers in women, so its diagnosis and monitoring is of paramount importance to society. Widespread adoption of tactile breast imaging would create significant impact, improving the way breast cancer is managed in the UK.

Tactile breast imaging is more comfortable for the patient than mammography, can image the region towards the shoulder that is missed by mammography and does not subject the patient to ionising radiation. Tactile sensing can image women with dense breast tissue (which is a limitation of Mammography), enabling it to image younger women. De-skilling the tactile imaging procedure through the creation of a global breast map, will pave the way for wider adoption of the technology, even to the point of home use, where high risk patients can perform self-examinations.

Ultimately, the research will enable the creation of a new medical device capable of producing a global stress image of a breast. Sales of such a medical device will generate revenue for UK industry and licensing income to the University. Tactile breast imaging is up to ten times more cost effective than alternatives such as mammography and considerably more portable. The cost, portability and ease of use of tactile imaging systems mean that they could be installed in a primary care setting, such as at GP practices, something which is not practical with other breast imaging systems. Adoption of a tactile breast imaging system could represent a substantial cost saving for the NHS. In particular, increasing the accuracy of GP breast cancer referrals from general practitioners by quickly diagnosing non-cancerous lumps (such as Cysts, Fibroadenomas and Diffuse nodularity) would cut costs and reduce patient waiting times at secondary care. See *1 for an overview of current procedures. Furthermore, there are applications in the underpinning research that would extend beyond tactile imaging to other medical imaging techniques such as ultrasonics, where the ability to accurately register the probe position would enhance the imaging process.

It is clear that this project is multi-disciplinary and that the outputs will be of interest to both the medical science and engineering communities. It is anticipated that cross-fertilisation between the fundamental science and the realisation of a prototype system will be the catalyst to explore new avenues of research within this exciting thematic area. Of particular note is the PI's parallel stream of research into structural imaging, such as pipework and vessels. There will be synergy with his other projects considering photogrammetry and visual inspection during manufacturing (EP/N018427/1) and for pipe inspection (Innovate UK CR&D Project Mosaicing for Automated Pipe Scanning - 102067).

The PI will pursue the conventional impact and dissemination activities through academic journal and conference publications. The proposed dissemination activities will allow the research to have strong exposure in the medical imaging field and beyond, leading to new knowledge exchange and joint research activities funded through the current Horizon 2020 framework.

*1 http://www.cancerresearchuk.org/about-cancer/type/breast-cancer/about/should-i-see-a-breast-cancer-specialist

Publications

10 25 50
 
Description We have learned how to make paupable reasonable breast phantoms with a vein network that is visible to an IR camera. Work to characterise these is underway.

We have a better understanding of the breast deformation likely in the population.

We have some preliminary results comparing featured based image matching to direct matching algorithms. We have some some surface based reconstructions, but they are at an early stage.
Exploitation Route As above
Sectors Healthcare,Other

 
Title Data for: "Improvement of Breast Cancer Tactile Imaging through Non-Rigid Mosaicing" 
Description Camera images contains bitmap images. Vicon path contains the movement of the camera (X,Y,Z) with 6-degree freedom. 
Type Of Material Database/Collection of data 
Year Produced 2017 
Provided To Others? Yes  
Impact Unknown