Digital Pathology is the process of converting glass slides into digital images to augment pathological diagnosis. The dynamic process enables pathologists improve efficiency and collaborate with consistency and transparency.
However, considering the current status of restricted movement due to pandemic, in 2020, the AI enabled digital pathology market may have grown slower than anticipated, yet the demand for online consultation and diagnosis will ramp up the demand from 2021 onwards. Young pathologists are preferring more online methods than actual lab visits though senior professionals are yet sceptical about the switch.
Technology advancements do come with challenges. This article discusses how artificial intelligence is enabling pathologists in discovering diseases with more precision and speed.
The role of AI in digital pathology
Though digital pathology aims to help pathologists spot subtle patterns and gain detailed information quickly, it also means that pathologists must deal with the large quantities of data created by digital pathology images. With AI, pathologists can overcome this challenge effectively.
Here’s a brief description to understand digital pathology and the role of AI in the process. Consider a typical digital pathology test where body tissues are examined using image analysis. During the analysis, an updated digital workflow is used to process the slides. Once processed, these slides are scanned and uploaded into software programs that use machine learning to identify patterns and provide information to pathologists.
The convergence of advanced imaging and enabling technologies like AI, machine learning, deep learning, NLP, data science, etc. is empowering the healthcare industry to unlock medical breakthroughs at an unprecedented pace.
Companies developing AI to automate the monotonous and tedious aspects of pathology
Following is a list of companies that are developing AI in digital pathology:
- Paige: Based out of New York, Paige is a software company that provides AI tools to help pathologists and clinicians make faster and more informed diagnosis. A spin-out from Memorial Sloan Kettering Cancer Center, Paige, uses ML with Convolutional Neural Networks (CNN) to better map the pathology of cancer to understand the origin and progress of the disease.
- PathAI: A spinoff from Beth Israel Deaconess Medical Center, PathAI is an MA-based company that uses ML with CNN to build diagnostic algorithms. PathAI has recently secured over $11M in Series A funding and is collaborating with Novartis to decode cancer pathology images.
- Proscia: Based in Philadelphia, PA, Proscia is a pathology AI company that is leveraging the power of data-centric medicine to combat cancer. Advanced Pathology Associates in suburban Maryland uses Proscia’s digital workflow platform for clinical uses cases. Also, Proscia is collaborating with Dermpath Lab of Central States (DLCS) to build diagnostic AI tools to detect skin pathologies.
- Google AI Healthcare: One of the well-known tech giants, Google, posits that AI is the future of healthcare. A deep learning algorithm that was recently developed by Google can identify signs of diabetic retinopathy from retinal scans with over 90% accuracy. In pathology, Google AI had recently participated in a deep learning study associated with prostate cancer and produced more accurate scores in Gleason grading than a generalist pathologist.
- Indica Labs: A leading provider of computational pathology, Indica Labs provides quantitative digital pathology services including immunotherapy predictive analysis, multiplex IHC, image management system, etc. Indica also has a deep learning algorithm called HALO AI that enables pathologists to operate remotely. HALO AI gained popularity in the Camelyon 17 Challenge for its impressive performance in identifying metastatic breast cancer.
How digital pathology is connected to patient care
Digital pathology has the potential to change the quality as well as the speed of patient care forever. For instance, before digitization, pathologists could look at only one slide under a conventional microscope, in which only a particular part of the entire tissue sample could be focused. Even the elite pathologists can miss things when studying the samples in fragmented views. Digital pathology puts an end to this challenge as it enables pathologists to view the whole slide as a single picture and zoom into areas of interest.
Also, many images can be viewed side by side, which means multiple pictures of a particular tissue sample can be examined at the same time and more granularly. This consequently helps in improving the accuracy and speed of diagnosis, thereby providing prompt and more quality patient care. Also, digital pathology supports information sharing, making it even more easy for clinicians to get second opinions from their counterparts.
Chief Innovation Officer Global Healthcare & Life Sciences at Dell Technologies, David Dimond, says that “In digital pathology, AI acts as a validation tool in imaging analytics and helps pathologists process more slides in a short time. Also, AI in digital pathology can impact patient experience profoundly, as patients can directly access their EHR (electronic health records) through mobile devices and applications.”
Though AI promises amazing advantages in digital pathology (increased efficiency, greater diagnostic accuracy, improved patient care, among others), the technology advancement could be initially cost-intensive as algorithm development and establishing the appropriate IT infrastructure may drive up costs.
Benefits of Artificial Intelligence in digital pathology
Following are some of the applications that exhibit the collective power of digital pathology and AI:
- More personalized treatments: AI and digital pathology are considered as an effective combination by most physicians as the two enable them to provide more patient-centric treatments. To get a clear perspective, consider cancer treatment – there are hundreds of possible diagnoses for a single type of cancer. According to estimates, there are nearly over 120 types of cancer and hereditary syndromes related to each. This a massive challenge for oncologists. AI and digital pathology enable healthcare professionals to make quick diagnoses and more informed decisions regarding a patient’s treatment plan.
- Improved efficiency: One of the most useful developments of the digital age is Electronic Health Records (EHR). Through EHR platforms, physicians can share information in real-time and provide better patient care. With the help of artificial intelligence, large volumes of datasets can be quickly analysed, and thereby the efficiency of pathologists and physicians can be greatly improved.
- Optimize workflows: Lymph Node Assistant (LYNA) – a recent technology innovation of Google has gained the attention of people working in digital pathology. According to Google’s report, 99% of the time, LYNA can successfully detect metastatic breast cancer on slides. Pathologists who have used LYNA posit that the deep learning AI provides them time-saving benefits in various tasks, particularly the labour-intensive ones and enables them to spend more time on other challenging tasks. For example, with LYNA the average slide review time was reduced to half when analysing small metastases.
- More time to address abnormal cases: The two key problems that many pathology departments face are: One – the department is understaffed, two – the vast amount of data that is generated due to digital pathology could exacerbate the former problem. Also, many physicians claim that most biopsies that they see are normal and that they prefer to dedicate more time to abnormal cases. According to an American gastrointestinal pathologist – over half of the endoscopic biopsies are problem-free. AI can identify cases that are normal and flag the ones that need further investigation by pathologists (Irish Times).
So, will AI replace pathologists or support them further?
While digital pathology has made life easier for pathologists, AI is supercharging them by automating monotonous, tedious, and time-taking manual tasks. Besides enabling pathologists to analyse more samples, AI is also assisting them to find more in the samples. In short, the powerful AI systems can surpass human capabilities, thereby inducing fear of replacement among pathologists. So, should pathologists be concerned about AI?
The answer is No. AI is here to assist pathologists and make them better in what they do. With AI, pathologists can dedicate their time to more meaningful tasks and be more efficient in their jobs.
Before the Artificial Brains Takeover
In the coming years, precise genetic testing will be available at an affordable price and people will have increased access to precision medicine solutions like digital pathology. People will no longer consult specialists for an initial diagnosis – rather they will upload images to the cloud and have an algorithm do the initial diagnosis. Also, AI enables pathologists and other healthcare professionals to be confident decision makers. Though challenges are likely to crop up while embracing the digital transformation, the applications of AI in digital pathology is paramount and worth the hardships.
