CM-Path at the NCRI Conference

Find out about what CM-Path members heard and the activities they took part in over the NCRI conference 2018 in Glasgow.

Report from the NCRI Conference Digital Pathology session

Author: Hayley Morris, CM-Path Workstream 1 member

This session, chaired by Dr. Karin Oien, University of Glasgow, presented a selection of talks on the use of artificial intelligence and deep learning in digital pathology. Dr. Oien introduced the session, outlining the need to embrace new developments in digital pathology. Giving the example of a recent survey by the Royal College of Pathologists which found that most NHS Histopathology Departments do not believe that they currently have sufficient staff to meet clinical demand, she argued that we should regard these as exciting opportunities to assist, rather than compete with, pathologists. The recently published RCPath document, “Best practice recommendations for implementing digital pathology” (https://www.rcpath.org/profession/digital-pathology.html) is a positive indication that the profession is working towards integration of digital pathology into routine clinical practice.

Professor Manuel Salto-Tellez of Queen’s University, Belfast gave the first talk of the session, in which he proposed that artificial intelligence will represent the third major revolution in pathology (following the emergence of immunohistochemistry and molecular pathology as the first and second revolutions). He then went on to discuss some of the projects in which he has been involved including the development of deep learning tools which help to identify regions of interest in tissue samples for molecular testing (TissueMark) and open source software for digital pathology analysis including biomarker quantification (QuPath).

Many of the projects discussed during the session were excellent examples of collaboration, both between academic centres and with industry partners. One such examples is the PathLAKE (Pathology image data Lake for Analytics, Knowledge and Education) consortium, which Queen’s University has joined and which has just received a large investment from the UK Research and Innovation Industrial Strategy Challenge Fund to advance the use of artificial intelligence in diagnosis and precision medicine. Professor Salto-Tellez introduced the consortium, before handing over to the next speaker, Professor Nasir Rajpoot, University of Warwick, who is leading the computational arm of the project.

Professor Rajpoot explained that from the perspective of a data scientist, histological slides are very data rich and therefore an interesting focus of research. He discussed how human pathologists and artificial intelligence can complement one another – while computers are very good at counting objects and intensities accurately, human brains can integrate a lot of different kinds of data, make inferences from this and draw clinically relevant conclusions in a way computers cannot. He argued that utilising the skills of both will improve the quality of data generated and therefore the output of the report. It was interesting to hear of some of the projects he has been previously involved in, which included developing algorithms to normalise staining intensity, detection of lymph node metastases in breast cancer and profiling of the tumour microenvironment.

Finally, Dr. Yinyin Yuan, leader of the Computational Pathology and Integrative Genomics team, Institute of Cancer Research, London, discussed her work on digital analysis of the tumour microenvironment, including development of algorithms for automated immune spatial scoring in ER-positive breast cancer, multiplex immunohistochemistry to assess heterogeneity of the immune infiltrate within non small cell lung carcinoma and variation in cell morphology within ovarian cancer.

It was interesting to observe the diversity of questions being investigated with artificial intelligence approaches. The audience were clearly engaged in the session and this was reflected in a very interactive question and answer session with the panel, with the overall tone being of positivity about the future of research in this area. With established multi-centre collaborations and the recognition of the importance of pursuing research in this area and translating it into clinical practice, as evidenced by the major investment from the Industrial Strategy Challenge Fund, digital pathology promises an exciting future!