Micrima appoints leading AI scientist to accelerate its product development
Christian Graff, PhD has emigrated to the UK to join the development team at Micrima Ltd to work with Artificial Intelligence (AI) algorithms that should aid better breast cancer detection.
The Canadian Scientist previously worked in the USA for the Food and Drugs Administration (FDA) where he was one of the FDA’s primary technical experts for machine learning technologies applied to medical imaging systems. With considerable experience both in research related to performance evaluation, and in assisting companies, from small start-ups to large multinational corporations, through the process of transitioning machine learning technologies into validated commercial products, Christian’s knowledge and skills will be invaluable.
The Micrima technology that Christian will be working with ‘MARIA®’ (Multistatic, Array Radio-wave Image Acquisition), uses harmless radio-waves to create 3-dimensional breast volumes. The technology originally developed by the university of Bristol, works on the principle that cancerous cells have very different dielectric properties to those of healthy breast tissue. With conventional imaging such as mammography, radiologists must visually discriminate very small differences in visual contrast, with factors such breast density often masking what they need to see. With the dielectric contrast between cancerous cells and healthy breast tissue being as much as 10-1 (Phys Med Biol. 2007 Oct 21;52(20):6093-115 Lazebnik M et al) radio-waves can detect these significant differences as well as provide a multitude of additional information.
The MARIA® technology is currently undergoing clinical trials at the Royal Marsden hospital in Chelsea, London. At present radiologists purely report on the intensity of signal reflected back from cells within the breast, but the work being undertaken by Christian and the other developers within Micrima should add additional functionality to this. Micrima’s current AI development focuses around the detection of known anatomical markers, such as the nipple and chest wall, then classifies lesion related findings against previous pathology that has been imaged with this technology vs the results obtained when those previous patients were biopsied. It is hoped that the technology will not only provide a comfortable (no breast compression), radiation free imaging solution that is ideal for dense tissue but will also increase cancer detection and reduce unnecessary biopsy’s where the technology indicates a finding is likely to be benign.