D I G I TA L PAT H O LO G Y 1 5 Radiomics, pathomics and deep learning Computational imaging furthers precision medicine The advent of digital pathology provides the means to develop computerised image analysis to diagnose disease and predict out- comes for cancer patients from histopathology tissue sections. Report: Mark Nicholls Such advances can help predict the risk of recurrence, disease aggres- siveness and long-term surviv- al, according to leading expert Professor Anant Madabhushi, of Case Western Reserve University, Ohio, USA. Speaking to online delegates at the 7th Digital Pathology and AI Congress, Madabhushi outlined how tools are being developed to advance this area of diagnosis, prognostics and prediction. His team has developed a suite of image processing, computer vision and machine learning tools spe- cifically designed to predict disease progression and response to therapy via the extraction and analysis of image-based histological biomarkers derived from digitised tissue biopsy specimens. These have already been applied in the context of several different disease domains including breast, lung, oropharyngeal, prostate, ovarian cancer, and endomyocardial biopsies. He suggested the tools would serve as an attractive and cost-effec- tive alternative to molecular based assays, which attempt to perform the same predictions. His presenta- tion, ‘Computational Pathology as a Companion Diagnostic: Implications for Precision Medicine’, looked at the implications of such tools for preci- sion medicine against a backdrop of cancer diagnosis and mortality in the USA. With 600,000 deaths annually Spatial distribution of prostate cancer across a cohort of 80 men, colors reflect the frequency of occurrence in different prostate zones said, but they are expensive and have a limited success rate of 20-25%, and no clarity on which patients will benefit or respond to these therapies. This, he pointed out, underlines the need for better diagnostic, prog- nostic and predictive tools that will identify the presence of disease, as well as predict disease outcome and progression and the response to treatment. ‘AI, deep learning and machine learning can really aid the pathologist, but there is a real unmet clinical need for these tools which can benefit the clinician, the one who is interfacing with the patient and 40% of the population receiving a cancer diagnosis, he noted: ‘There is some discrepancy between those two statistics. ‘Some of the reasons why diagnos- tic incidence is high and mortality for the disease not in the same ball park is because we have become more aggressive about cancer detection, screening and imaging.’ Overdiagnosis harms patients Madabhushi warned that overdiag- nosis is harming patients – not only physically due to the treatments and side effects of therapy, but also financially with a high proportion of cancer patients having to invest their life savings in their treatment. While finding disease earlier leads to a more favourable outcome, he suggested computerised image analy- sis could play an increasingly impor- tant role in the cancer diagnosis and therapy evaluation. Immunotherapy is a game-changer for cancers such as melanoma and lung cancer, he and having to make treatment and management decisions. ‘This will support precision medi- cine by using prognostic and predic- tive tools for tailoring therapy for a given patient based on a specific risk profile.’ His group is working to under- stand the full value to be gained from the data gleaned from tissue biopsies and the unprecedented opportunity to use computational machine learn- ing tools on the digitised slides. ‘This will not only identify presence or absence of disease,’ he said, ‘but also mine data for digital interroga- tion of data from the perspective of mining digital biomarkers, to iden- tify features that can tell us about the risk of progression of disease, aggressiveness of disease, and how likely patients are going to respond to chemotherapy or immunotherapy.’ this illustrate Studies prove benefit To potential, Madabhushi highlighted a series of studies from his group where AI and machine learning had been shown to suggest which patients would benefit most from which treatment. These included one focusing on disorder of collagen fibre associated with risk of recurrence in Oestrogen Receptor Positive (ER+) breast can- cers in ECOG-ACRIN E2197 & TCGA. ‘This goes to show the value of these diagnostic tools to aid the clinician in figuring what the outcome is and how to treat these patients,’ he confirmed. ‘This is also true in early- stage disease.’ In other examples, Madabhushi outlined how a combination of com- puter extracted features of nuclear morphology, tubular formation and mitotic count predict disease-free Digitisation dawns in developing world The future of digital pathology is assured Pathologist Dr Talat Zehra reports from Pakistan Given the rapid transi- tion towards digitisa- tion, digital pathology is now unquestionably the future. However, some pathologists, particu- larly in underdevel- oped countries, are still reluctant to accept its place in their labs. Among their many reasons, some feel that histopathology is a very complex and subjective field and artificial intelligence (AI) software cannot cope with all the issues. Conversely, pathologists are also scared that AI might replace them completely. Last but not least, a large number of pathologists, particularly senior ones, or those who have not worked in the developed world, are not familiar with new modalities, so they are reluctant to adopt them. www.healthcare-in-europe.com Pathologists in many countries are eager to go fully digital – however, in some cases, crucial pieces are still missing (Image courtesy of Dr Alex Wright, University of Leeds) As for Pakistan, the bottleneck here is either the absence of pathology slide scanners, digital microscopes or manual whole slide image (WSI) software; if available ,they are main- ly used for educational or research purposes only. The use of AI is almost negligible, indeed very few pathologists know about this novel entity. Initially, I did some pilot projects to validate the results of AI software on previously diagnosed cases. For this I am highly thankful to Aiforia Technologies Oy, which gave me its demonstration version and training on AI software. Using this, I carried out some pilot projects on chorionic villi and malarial parasite detection and gained a good concordance of around 84%. We accomplished this project without a scanner or digital microscope. Despite slow adoption in many institutions and countries, the digital slide image is slowly replacing the glass slide. Aided by AI-based image analysis software, we can improve the much weaker and fragile health- care delivery in developing coun- Anant Madabhushi is the Donnell Institute Professor of Biomedical Engineering, Case School of Engineering, Case Western Reserve University, Cleveland, Ohio, and also Director of the Centre for Computational Imaging and Personalised Diagnostics. In the School of Medicine he is Professor in the Departments of Pathology, Radiation Oncology, Radiology, Urology and General Medical Sciences. His research interests include quantitative image analysis, radi- omics, pathomics (a central source for computational pathology), cancer imaging, and digital pathology. survival in ER+ breast cancer, and also how computer extracted images of nuclear shape, orientation disor- der and texture from whole slide imaging are associated with disease- free survival in ductal carcinoma in situ. He further detailed possibility of using similar approaches predicting post-surgical recurrence in prostate cancer, lung cancer, ovarian cancer, endometrial and cervical cancer. ‘Computational analysis with rou- tine imaging could help address questions in precision medicine, spe- cifically prognosis and predicting response to therapy,’ Madabhushi concluded. ‘The relatively low-cost aspect of computational diagnostic tools could also have global benefits, especially in low- and middle-income countries.’ And, he predicted, ‘the importance of establishing the con- nection to the molecular underpin- ning of the features, moving away from an abstract representation, is going to be a really important driver for clinical adoption and clinical utility.’ Dr Talat Zehra MBBS FCPS is assis- tant professor at Jinnah Sindh Medical University (JSMU) and consultant Histo- pathologist at JSMU Diagnostic Lab, Karachi, Pakistan. She gained her Bachelor of Medicine, Bachelor of Surgery degree in 2007 from Dow Medical College, Uni- versity of Karachi, Pakistan, and her fel- lowship in Histopathology at the College of Physicians and Surgeons of Pakistan. Zehra’s field of interest is digital pathology and the use of artificial intelligence in tis- sue imaging. Zehra has written a few arti- cles internationally to highlight the issues in delaying adoption of digital pathol- ogy techniques in the developing world. Currently, she is also a member of the edu- cation committee of the Digital Pathology Association (DPA). tries. These carry most of the world’s endemic disease workload but, Continued on page 16 Ki67, ER, PR, Her2 Imaging•Fully automatic•For WSI or microscope camera•LIMS IntegratedDigital Pathology Solutionswww.vmscope.com