About the project
A proprietary technology for healthcare
AI is a powerful force disrupting the medical industry and its many workflows. So powerful in fact, that it brings us a step closer to beating one of the world’s ultimate challenges – cancer. To aid this race, Tooploox (now part of Solvd), brings Virtum, the AI-ready digital pathology workflow platform.
The Virtum prototype was initially delivered by MicroscopeIT, which was later acquired by Tooploox to enhance its medtech solutions portfolio and support the technology’s development. Tooploox subsequently joined Solvd, continuing the development of Virtum within a broader healthcare and AI innovation ecosystem.
Virtum is a comprehensive platform for automated image analysis and an enabler for clinical AI solutions. It emerged from the need to support, enhance and automate histopathology and medical imaging workflows. Virtum is an outcome of a collaboration between histopathologists, scientists and software engineers. With histopathology as the top-of-mind use case, Virtum is a universal workflow collaboration and storage system aimed at healthcare professionals, with the goal of building an environment for machine learning in medical imaging. This deep understanding of the domain, combined with technical proficiency, resulted in a tool that addresses key challenges in the medical imaging and histopathology fields.
The challenge
Images so detailed that it is impossible to share
Medical imaging and histopathology today
The WHO data shows that the leading causes of global deaths are non-communicable diseases, with cancer being one of the most common. Also, up to 50% of cancer cases can be avoided with relative ease by avoiding the risk factors. The next most effective way to reduce the number of cancer-related deaths is early detection.
The early diagnosis can be provided by regular check-ups and access to modern healthcare advanced imaging devices like CT scanners or MRI as well as by skilled teams of medical imaging specialists and histopathologists alike.
What is a medical imaging?
Medical Imaging is an umbrella term describing the process of taking images
of the internal human body as well as analyzing the functions and flow of particular organs and tissues.
It is done using devices like ultrasound scanners, Magnetic Resonance Imaging (MRI), computer tomography (CT) or X-ray scanners, among others, producing various types of medical scans.
Medical imaging is a way to spot suspicious changes in the body. If the organ or tissue
is removed, it is further analyzed within the pathology workflows, usually histopathology.
What is histopathology?
Histopathology focuses on examining the tissue on a microscopic level. While medical imaging can determine if there is some suspicious change in the body, histopathologists deliver
a detailed diagnosis on the type of the detected change.
IBM researchers estimate that up
to 90% of modern healthcare data consists of images, with medical imaging and histopathology contributing significantly to this volume. While specialists from both areas save lives on a daily basis using their sophisticated equipment, there are several challenges to overcome.
While the new diagnostic tools bring new possibilities to modern healthcare, there are also multiple challenges to overcome to get the full benefit out of them.
According to IBM data, up to 90% of modern healthcare data consists of images, including MRI scans, X-rays, histopathology images and many other types, depending on the stage and procedure. The more detailed the image is during the early diagnostic process, the more likely it is to spot early signs of a disease, when it may be easier to manage or even entirely prevent. Later, these images are used to control the treatment process. Better quality may help identify suspicious or early changes in tissue or subtle changes in the patient’s condition, allowing medical professionals to initiate treatments earlier, ultimately leading to better recovery chances. The bigger and more detailed the image, the more accurate the information about the patient is. The problem with this type of data is that it is challenging to process and share – a single image can be immense.
Challenges with sharing make consultations complicated as healthcare professionals cannot simply send a scan to someone for consultation, even if it is necessary for a particularly complex diagnosis. This is further compounded by a shrinking number of pathology specialists worldwide: according to the US National Library of Medicine, pathologists decreased from 2.03% of total US physicians in 2007 to 1.43% in 2017. Furthermore, medical tools are not AI-ready, fragmented, non-standardized and challenging to connect. By that, launching AI tools on medical imagery data poses an even greater challenge.
Virtum aimed to provide a unified, standardized and convenient environment for working with medical imaging data. The goal of the creators was to deliver “Google Docs for medical images,” where one could share data with peers, store it, highlight specific parts and perform all the tasks known from modern, sophisticated office suites in a compliant and modern environment. Moreover, it was also viewed as a way to annotate the data and to share a suspicious part with peers, to get a second or third opinion on the matter.
Technologies and partners
Popular technologies for easy development and modification
The outcome
Modern platform for better healthcare
The deep understanding of the context, combined with technical proficiency, resulted in delivering a tool that has addressed the main challenges in the medical imaging and histopathology fields.
Searchable and legible sample archive, quick access to any image
Contrary to more traditional methods of storing samples, these are fully searchable and can be tagged or marked in other ways to make them easy to find in the future. Over the course of many projects, Virtum had been adopted for work on High Performance Computing (HPC) clusters, for instance, Fortissimo.
All-in-one workflow platform, tool-juggling no more
Virtum provides the tools to store, manage, analyze and work on microscopic images. There is no need to move the image outside the secure Virtum environment; it meets the security standards hospitals and healthcare institutions require. The platform also includes a pyramidal data viewer that enables users to smoothly display and process even large and heavy images, such as histopathological scans or MRI images. For annotation, Virtum offers a rich selection of image-editing tools, including a magic wand tool to select pixels that share the same color and brushes of multiple sizes.
Easy collaboration on any image, consult with peers from all around the world
Virtum learns from the experiences of Google Workspace and similar collaboration tools, bringing the ease of sharing, commenting and editing to digital oncology and digital pathology systems. Users can check the version history, reply to comments in threads, share images and withdraw access at will.
This presents an excellent opportunity for healthcare professionals to consult with peers from around the world on their cases. The user to whom the image was shared needs only a browser to access the full benefits of using Virtum and can support their colleagues in providing better care for patients.
Cloud or on-prem infrastructure to better address the compliance
According to GDPR, it is often preferable for institutions to keep all data in-house. Technologies used to build Virtum support the on-prem infrastructure, leaving no uncertainty regarding compliance. As a cloud-based solution, Virtum also provides access to computing power that scales up when needed and scales down when not required.
AI-ready for both labelling and operations, boost your microscopic workflows with AI
Virtum is accessible via API, making the tool ideal for Artificial Intelligence applications. Teams worldwide can utilize Virtum for data labeling to create an AI-ready dataset, training the algorithm to support their workflow.
The algorithm can be further implemented within the Virtum workflow by allowing it to function as a separate component that connects with the rest of the suite via API. There is also no obstacle to using Virtum as a platform to run the AI algorithm and send the effects to other tools or workflows.
Virtum is also powerful enough to combine artificial intelligence and radiology. Being a format-agnostic tool, it can deliver AI image diagnostics for radiology and serve as a platform for AI in pathology solutions.
In the traditional workflow, users had to manually search through research papers on histopathology to compare an encountered anomaly with existing cases. Virtum, powered by AI algorithms, enables users to run an automated search through a database of research papers, looking for comparable cases, significantly enhancing the accuracy of diagnosis, even for extremely rare cases.
Originally, the tool was developed to support histopathologists. Adaptation of Virtum contributed to more efficient data modeling and helped enhance medical processes; however, it has also found applications in non-medical industries, including manufacturing, renewable energy, life sciences, biotech, space, mining and chemicals. The experiments involved microscope equipment manufacturers and medical R&D units to analyze images from remotely controlled microscopes, process super-resolution images, and take full advantage of the virtual microscopy toolbox.
