Computing Collaborative Research and Development with TenD Innovations Reaches an Important Milestone
6 June 2019
Our research partnership with Ten-D Innovations has recently reached an important milestone. The research team in medical image analysis has developed an effective deep learning approach that enables recognition of thyroid and breast cancers from ultrasound images with levels of accuracy that is significantly higher than that made by non-specialist doctors in clinics. The resulting recognition models for both types of cancers have been embedded into an automatic cancer detection software that was announced at a press release ceremony in Shanghai on 20th May 2019. The Buckingham delegation consisting of Dr Harin Sellahewa, Prof Sabah Jassim, Mr Hongbo Du and Dr Alaa Al Zoubi, headed by Dr John McIntosh CBE, Vice-chair of Council was present at the ceremony. The ceremony was attended by more than 200 people and received a wide coverage by the local and national press in China. All members of the delegation were interviewed by the Press in Shanghai. Mr Du and Dr Al Zoubi joint hands with Dr Yi-Cheng Zhu of Renmin Hospital in demonstrating the use of the software to the press. Prof Jassim and Mr Du also presented at a discussion panel after the ceremony. The software is currently on trial in several hospitals in Shanghai. TenD Innovations intends to roll out a commercial version of the software in the near future.
Pictured above: Mr Hongbo Du, Dr Alaa Al Zoubi and Dr Yi-Cheng Zhu of Renmin Hospital Shanghai demonstrating the software at the Press Release ceremony, Shanghai, 20 May 2019
At the ceremony, Dr John McIntosh CBE expressed his satisfaction that “the close collaboration between Buckingham and TenD Innovations has borne fruits in such a short period of time, thanks to the enormous hard work put into the research and development by both parties”. Dr Harin Sellahewa, interim Dean of School, added that “the partnership with Ten-D will enable us to harness the power of Artificial Intelligence to revolutionise healthcare. It exemplifies the University’s commitment to deepen collaborative relationships with businesses in the UK and overseas to address some of the key grand challenges of 21st century.” Mr Hongbo Du, the Centre Coordinator, also added that “our collaboration with TenD Innovations goes from strength to strength. Through the hard work of our research team, we have delivered good solutions at this initial stage. We will deepen as well as broaden our research into deep learning paradigms and develop more effective solutions in cancer recognition and detection to improve quality of life for the society”.
Ultrasound imaging is considered as a safe and non-intrusive image modality with no side-effect, and has been widely used in health check centres, clinics and hospitals. Medical diagnosis using ultrasound images alone is a difficult task; medical staff with different levels of training and experiences may come to different conclusions when examining an image. The success rate for correct diagnosis by inexperience doctors and radiologists is very limited. A reliable computer-aided diagnosis tool can provide doctors not only with a valuable second opinion but also a good training opportunity to improve their diagnostic skills. However, computer-based ultrasound image analysis is considered technically challenging due to poor image quality and presence of various noise signals in the image. The real-time dynamic process of image acquisition also add another factor of challenge and complexity. In the past several years, School of Computing has accumulated a huge amount of knowledge and research experience in this field. Currently, the TenD medical image analysis project involves 1 research fellow, 4 PhD research students, and members of academic staff within the School. Each research student investigates a specific aspect of deep learning for cancer recognition and detection.
Pictured above: Mr Hongbo Du and Prof Sabah Jassim appearing as Panellists at the Press Release ceremony, Shanghai, 20 May 2019