Artificial intelligence, AI, has been one of the most talked-about topics in pathology for years. AI stands on the shoulders of digital pathology, which can already feel threatening to the expert who has maybe spent decades looking at specimens in a microscope. Will all this rapidly developing technology make the expert obsolete?
Every year we know more and more about diseases and the intricacies of the human body, and healthcare professionals everywhere order more and more tests on their patients. At the same time people everywhere live longer, which brings along a growing number of cancer cases. Still, ever fewer young doctors specialize in pathology. This all inevitably means there just aren’t enough pathologists available for the increasing workload. To make matters worse, this is a global problem.
In conventional pathology a trained expert is present when the surgeon takes a biopsy of the patient, and can examine the samples right away. If there is no pathologists available onsite, the samples are sent to them, which may take days or even weeks depending on distance, method of transportation, and whether the case needs specific expertise. If the pathologist needs a second opinion for a difficult case, the samples will have to be sent off a second time.
Digital pathology is the obvious solution – samples are photographed with a high-precision microscope scanner, and instead of shipping slides, the image files are shared online. The expert pathologist can sit on the other side of the world and examine and diagnose or get a second opinion online at once. Unlike a unique physical sample, a scanned slide can also be viewed by almost any number of pathologists at the same time, which is excellent for educational purposes.
MACHINE VISION VS. ARTIFICIAL INTELLIGENCE DIAGNOSTICS
An increasing amount of work means the pathologist spends much of their time looking at samples on glass slides, browsing through one slide after another. Humans are not very consistent in mechanical, repetitive tasks which still require constant vigilance, and with a growing workload oversight and mistakes can happen. If the samples are scanned into digital files, artificial intelligence can be trained to filter out the obvious cases and only leave the ones requiring a trained eye to the human pathologist. A computer programmed to sort through image files is an example of machine vision. It is a practical tool, making the work of the human expert less tedious and monotonic, letting them focus on the interesting images and applying their experience and expertise in their work where most needed.
Computer vision is as good as the algorithm humans programmed for it. When the computer reading image after image is programmed to learn from the mass of image data it processes, it is called deep learning, a form of artificial intelligence. It is the computer mimicking a human expert’s training and learning. Teaching the computer to know what to look for in a tissue or fluid sample can take months or even years, and to make it accurate takes even longer. The number of diseases and different diagnoses in all the different disciplines of healthcare is huge, and even human experts are still far from understanding them all perfectly. Even if AI is a very talked-about subject and it holds a great promise, we’re still a long way from machines surpassing human expertise, let alone taking over.
DIGITAL TOOLS = BETTER HEALTHCARE FOR ALL
Artificial intelligence is not a threat to the human expert and won’t be any time soon. The use of digital tools and learning algorithms with machine vision will enable the human expert to use their time more sensibly: less repetitive tasks, more interesting work, which will in turn make the pathologist better at their work faster. The use of computers will increase consistency, reduce the number of mistakes, and lead to better acccuracy in diagnosis. Before long we can expect to be able to understand diseases better by looking at the mass of data the computers process ourselves, but also by following how the AI processes it and train AI algorithms accordingly.
The drawbacks in turning healthcare digital are in that it often takes a long time to implement. Making changes to a regulated workflow is usually not swift, nor always welcomed by those tasked with applying the changes. Few health systems employ innovator positions in their organization to find novel methods, let alone do research themselves. Innovation and progress then heavily relies on doctors who invest their time and interest in novel ways of healthcare as a pet project. Making changes can also be expensive, bigger change is more costly in both time spent on implementation and training, and actual new equipment and tools. It’s not unusual for a health system to be well aware of a solution to a problem, but they just can’t afford to make the switch for the time it takes and how much it costs. The Grundium Ocus microscope scanners have been designed with this exactly in mind in both the product itself and the business case.
IMAGING MADE PRACTICAL
The Grundium Ocus® is the most practical microscope scanner in the world. Its small footprint means it can find a spot in the most cluttered of labs. The Ocus® is so simple to use anybody can be trained to use it in just 15 minutes. Created by ex-Nokia flagship camera phone engineers, it takes the sharpest images in the business. It is much more affordable than big multislide scanners. Its software is so advanced it works with any existing laboratory information system (LIS) and it supports the most common file formats. The Ocus scanners simply remove the entry barrier to digital pathology.
If you are looking for a small footprint and sharp imaging component for your pathology solution or if you want to hear more about how to simply integrate an Ocus scanner into your system, please book a free, non-binding online demo of the Ocus scanners, or just message us with any questions.