Portugal’s IMP Diagnostics, a Digital Case Study
Family-owned IMP Diagnostics pioneered digital pathology integration into private practice workflows
Rob Monroe | | 10 min read | Discussion
This is the final story in a three-part series showcasing how pathology laboratories in New Zealand, Germany, and Portugal have successfully tackled the transition to digital. Read part 1 here and part 2 here.
Isabel Macedo Pinto first glimpsed the future of pathology more than 10 years ago.
“I was in a meeting and overheard two colleagues conferring on a complex case. I am intrigued by difficult cases, so I was drawn to their conversation,” says Macedo Pinto, who founded IMP Diagnostics more than 25 years ago and currently serves as co-CEO and Medical Director. “To my surprise, one colleague was attending our symposium in Barcelona and the other was in another country, yet they were reviewing an image together in real time. I immediately thought: “This is the future. We need to transform IMP to be digital.’”
Headquartered in Portugal, IMP Diagnostics is a private, family-owned, high-volume reference laboratory business with a central headquarters based in Porto and facility in Lisbon. More than 120 pathologists, lab technicians, cytotechnicians, and support team members provide comprehensive molecular and anatomic pathology services, accepting and reporting out cases from Angola, the Cape Verde Islands, Mozambique, and, of course, Portugal. IMP also has an R&D department that focuses on computational pathology.
João Sousa Pimentel oversees IMP’s strategy and technology, and serves as co-CEO and Head of Business Development. “IMP is structured for clinical excellence, efficiency, and growth. Subspecialized diagnostic units are the backbone of this structure, including gynecologic pathology, perinatal and placental pathology, breast pathology, gastrointestinal pathology, dermatopathology, and uropathology,” explains Sousa Pimentel. “We manage 250,000 cases annually, of which 135,000 are histology cases, 95,000 are cytology cases, and 20,000 are molecular pathology, with the number of molecular cases growing rapidly.”
The beginning of the journey – preparation to implementation
IMP began readying for digital pathology well before implementation. The process included planning the project’s feasibility, raising funds, and assembling the digital pathology team.
“We wanted to be sure fundamental elements were in place so we would be ready for full digital implementation when the time was right,” recalls Sousa Pimentel. For example, during the preparation phase, IMP integrated a 2D-barcode-based tracking system, which is critical for full digital pathology implementation.
“The digital pathology project team wasn’t big, but it was diverse. It included an overall project leader, a clinical lead to involve pathologists’ insights from the beginning, an IT representative to serve as point person with our tech vendors, and a business lead to oversee financial investments,” explains Sousa Pimentel.
As part of its diligence activities, the IMP team talked to hardware and software vendors and visited labs incorporating digital pathology. “I strongly recommend visiting with vendors and labs to see digital in everyday use. Doing so helped us gain an understanding of key considerations we would need to address, such as workflow modifications,” Macedo Pinto advises.
André Tavares, who leads IMP’s IT department, concurs with Macedo Pinto, and adds “Visits provide a true sense of a vendor’s experience working with digital pathology.”
IMP’s process enabled the team to identify pros, cons, and, importantly, remaining questions. “Three main questions emerged. The first was timing – when to ‘get in.’ The second question was how to manage the investment and choose the right solution. The final question was how to monetize digital pathology,” recounts Sousa Pimentel.
To help address the questions, the IMP team monitored digital pathology hardware and software development, regulatory actions, and publication of clinical evidence. Sousa Pimentel pinpoints three catalysts that, ultimately, compelled IMP to advance from a preparation phase to implementation.
The first catalyst was a 2017 presentation by Google CEO Sundar Pichai about machine learning in histopathology. “It was meaningful that the top person at Google took time at the company’s global developer meeting to address machine learning in pathology. We felt it was time to embrace digital or be left behind. So, we started looking even more closely at implementation,” says Sousa Pimentel.
The second catalyst involved two related, but distinct, events. “The year 2019 brought two defining moments in digital and computational pathology,” Sousa Pimentel continues. “A peer-reviewed paper was published documenting the performance of a clinical grade AI algorithm. That same year, the FDA granted breakthrough designation for the software referenced in that paper. The paper provided credible evidence of efficacy, while the FDA action underscored regulators’ belief that the software was ready for clinical use.”
Any lingering hesitancy at IMP was mitigated in 2020 with the regulatory approval and commercial launch of a first-in-class high-throughput scanner – the Aperio GT 450 DX. “The introduction of a ‘next generation’ scanner provided validation that hardware had caught up with software and that it would now be possible to bring digital into the lab workflow,” Sousa Pimentel recalls.
The halfway point – a stepwise approach
When the IMP team pivoted into an implementation phase, it was, naturally, not without its challenges.
“We are pioneers in the work to implement digital into private practice workflow – and we knew there would be challenges,” notes IT leader Tavares. “We are fortunate that everyone on the project team is truly committed to this endeavor, including our vendors. This level of partnership across the team is crucial.”
From the outset, Tavares and his IT colleagues had a goal to implement digital into IMP’s existing workflow without causing significant disruption to the pathologists. They knew IMP needed an approach to managing a huge amount of digital data in a manner that enabled the system to respond quickly and smoothly.
“Our aim was to have the digital system answer to pathologists in the same way an analog system would, and not interrupt or add time to the workflow,” says Tavares.
“Even a minimal change in workflow affects a lot of people,” notes João Monteiro, IMP’s Laboratory Director. “In the histology department alone, there are more than 25 technicians. Everyone should work the same way to get the best result, whether it’s cut thickness, slide disposition or organizing the workflow.”
To achieve this, IMP took a stepwise approach. The organization acquired the FDA-cleared high-throughput scanners, an action the team considers the ‘official’ kick-off of the implementation process. Scanner installation was followed by an in-house assessment of requirements for IT integration and initial vendor negotiation, with the LIS–IMS integration completed in six months. Concurrently, an initial test scanning phase was undertaken involving the digitization of 2963 slides (44 percent archive material and 56 percent current, routine cases), followed by the incremental digitization of the routine workload. IMP opted to begin by scanning single subspecialty areas and then scaled up to full digitization.
“The stepwise approach meant pathologists and the technical team could learn and adapt to digital over time, and that was central to our success,” says Diana Felizardo, a pathologist hired in 2020 to serve as IMP’s Clinical Lead of Digital Pathology and Unit Head of R&D.
“People accustomed to working with a microscope can be a little afraid of digital. We found the learning phase essential to creating an ease and familiarity with digital for our colleagues. Now, the pathologists report that they not only like digital but are also comfortable working with the technology. They don’t want to go back to the microscope because the microscope is not as practical in today’s work environment,” says Felizardo.
“Many pathologists’ initial reaction to digital is concern but, once they see that it works, they realize it can be a very good thing. We can do different – often better – tasks with digital pathology,” adds Felizardo.
IMP is documenting the impact of digital across the practice and finding that although digital pathology deployment in the private setting has its challenges, it also provides compelling opportunities, including easier and faster case delivery to the pathologists. This is extremely relevant for IMP, as they have two Porto buildings, as well as two laboratories in different cities, meaning that physical slides need to be transported between sites. Digital also allows simpler case sharing between colleagues located in different places and enables remote work.
“There will be hiccups, just as you would expect with any digital system. In the end, however, the benefits of digital outweigh not having it,” acknowledges Felizardo.
On the horizon – computational pathology enhances patient care
Digital pathology deployment lays the foundation for the integration of artificial intelligence (AI) tools, which will ultimately contribute to improving patient care.
“AI requires access to digitized pathology slides, which are generated using whole-slide scanning,” explains Vivian O. Tan, a computational pathologist with clinical diagnostics developer and supplier Leica Biosystems. “AI tools can be key in increasing workflow efficiency and improving diagnostic quality.”
AI applications in pathology are already in use in clinical and research settings. There are digital tools available to assist diagnosis that can facilitate many pathologists’ tasks, such as measurements and annotation of regions of interest. State-of-the-art AI approaches can be used for advanced tasks, including prediction of survival and therapy response, which, if rigorously validated, can enhance clinical decision-making in the future.
Tan further explains, “AI enables researchers to identify biomarkers that can be useful in the clinical setting to aid in treatment decisions. AI can identify morphologic features on H&E that correlate with molecular changes, for instance, that may be associated with drug benefits for specific cancers. All of this is made possible by AI. And digital pathology is the substrate for AI.”
Sousa Pimentel says, “At IMP, we think of digital pathology as a vessel. It can take us from point A – enabling diagnosis largely in the way pathologists currently work – to point B – diagnosis with AI.”
The organization is making this a reality with an ambitious digital pathology/AI agenda. IMP implemented digital pathology in parallel with computational pathology, starting with a research project to build algorithms for the detection of colorectal and cervical cancers in scanned images. “We aim to extract new types of information from scanned images to help better manage and even predict disease,” says Felizardo. IMP is also building a digestive cancer algorithm to aid pathologists in analyzing histopathology images of colorectal carcinoma and other GI cancers in a more accurate manner.
“Digital pathology has transformed our current diagnoses into more exact and complete forms of diagnoses, and it will continue to do so as more AI algorithms are developed,” reflects Macedo Pinto, who believes AI will both enable an economic return on investment and make pathologists even more vital to patient care.
She continues, “There is potential for an economical return in addition to the clinical return. If I can provide a diagnosis more rapidly and leverage AI to make the diagnosis more precise and comprehensive, colleagues will select IMP as their partner. They’re more likely to send us their biopsies and specimens because of the in-depth insights we can provide.”
Felizardo emphasizes this sentiment: “Every pathologist is a doctor first, with the aim to help patients. For those of us working in health research, we do so to find novel insights to improve patient care.”
Looking ahead to the next decade? Macedo Pinto says, “AI will enable us to better serve clinicians and their patients. It is our commitment to them that drives us at IMP to continue to innovate. We never forget that each slide we review represents a human life. For this reason, we seek to maximize the excellence of the diagnostic services we provide, every day, for every case. We need to be exhaustive. We need to be exact. We need to give these precious diagnoses to our patients.”
Image Credit: The Pathologist
Pathologist currently serving as Chief Medical Officer for Leica Biosystems and Chief Scientific Officer, Oncology, for Danaher Diagnostics. Monroe is board certified in cytopathology, anatomic pathology, and clinical pathology and holds a PhD in genetics. He has years of experience in the digital pathology space and frequently consults with pathologists around the world.
