To find out the latest news concerning ESP and its activities, please follow this Link to Members Area of the ESP website to read the Newsletter.

On 13 November 2024 we will celebrate the International Pathology Day. This date recognizes the contributions of pathology and laboratory medicine to global health and wellbeing.

Bulgarian Pathology Association publish an interview with Prof. Catarina Eloy from Autumn 2024 Edition of ESP


ESP Interview Series Interviewee: Prof. Catarina Eloy Head of Pathology Laboratory of IPATIMUP, Porto Interviewer: Prof. Marta C Cohen

Professor Catarina Eloy is an Affiliated Professor of the Medical Faculty of Porto University (Portugal) since 2015. She is an interventional pathologist, researcher, and teacher devoted to the diagnosis and study of thyroid cancer. She has published more than 100 manuscripts in indexed scientific journals and is the Head of the fully digital Pathology Laboratory of Ipatimup, Porto, since 2013. Prof. Eloy developed a special interest in the modernization of the diagnostic processes, including digital and computational pathology. She is the Chair of the Educational Subcommittee as well as Council Member of the ESP since 2023, Prof. Eloy is also member of the ESP Advisory Board, of the Editorial Board of Virchows Archiv and coordinator of the ESP Advanced Training Centre for Digital Pathology at IPATIMUP.

How do you believe digital pathology and AI will fundamentally change the way pathologists work in the near future? The digital transformation of pathology is an extension of the digital transformation observed in other services, namely in those related with health care, and will lead to an inevitable automation and robotization of our departments, ultimately transferring the diagnostic exercise to an in-silico environment. This process requires new competences and financial challenges and aims to achieve better sample management through standardization of the way we work. It is also an opportunity to address the necessary change in the concept of pathology as a more technological discipline, and reinforce the added value that pathology may carry to patient care. The observation of the sample in a computer screen, instead of using a microscope, is the step that allows us to have access to the computational pathology applications, namely those called AI tools.

What are the most significant benefits you have seen or foresee with the integration of AI into routine pathology practice? As for today, AI does not work alone, allowing us to go for a coffee while it reports on our behalf. Instead, AI tools are already contributing to making our work more efficient, less error prone (because these tools may highlight dispersed morphological events), and more precise in quantifying features, such as those related with biomarker’s evaluation. It has a lot to do with an increment in the comfort of our daily routine, but also brings new horizons in terms of better understanding and classifying dis eases.

In your opinion, what governance measures should be in place to ensure the responsible and ethical use of AI in pathology? I gladly recognize that we are seeing a progressive interest from regulatory institutions to assure the responsible use of AI in health care and in particular, in pathology. It is now clear that the use of AI in pathology should be carried out by trained pathologists, using verifiable tools, under a tight quality control program. The synergic model of operating, meaning to work together with AI as a tool and not as a substitute for our work, is the one that has demonstrated the greatest advantages and safety (since the pathologist has the final responsibility for the diagnosis).

What do you think are the main barriers preventing the widespread adoption of AI in pathology today? As in the case of other new technologies, the financial investment and reimbursement policies are at the top of the list of main barriers. Besides that, and probably more difficult to achieve dedicated funds, is the need for a solid and universal digital transformation of pathology departments that is far from happening. To start working with AI tools without a properly settled digital transformation able to generate standardized, high-quality, whole slide images for clinical use is a mistake that will inevitably lead to a decrease in our performance as diagnosticians.

What further developments in AI technology do you believe are necessary to fully realize its potential in pathology? Provided that we operate with AI tools on top of standardized, high-quality, whole slide images, we should only make sure that the selected software is adequate and safe. There are numerous proofs of concepts in the literature that state on the good performance of AI tools in controlled settings. Nevertheless, for most of the AI tools avail able and approved for clinical use, there are no clinical validation trails or prospective evaluations. In addition, taking into consideration all the variability in tissue processing happening in each institution, we also miss, for each AI product, the list of specifications for use as well as indications for proper control measures. In our experience, after using AI tools in the clinical set ting for almost 2 years, all AI tools require internal validation.

What kind of training or educational programs do you think pathologists need to effectively work with AI-based tools? This is a topic that is underestimated. If fact, the majority of the education initiatives on AI dedicated to pathology drive inevitably to basics on image analysis and training algorithms. It is true that pathologists need to have some basic instruction on this to choose wisely the best AI tool but what we need to know is how to use it without running into bias/deskilling and how to control it. The human-machine interaction in pathology has many risks and only those who live it are those who feel the need to build barriers to protect themselves from a progressively poor performance. There is definitely the need for education programs dedicated to this subject.