Title goes here
About us
Total 39 companies invested
2020
Total 79 companies invested
2021
AI in Medical Imaging: Challenges and Perspectives
In 2020, the COVID-19 pandemic put an unprecedented load on medical systems worldwide. While the lack of capacity had been felt earlier, the pandemic highlighted a severe deficiency. Developing technological instruments empowering medical staff to treat more patients attracted attention. Moreover, technological advancement and data availability have made the healthcare industry more data-driven and analytical. In contrast, the quality and the speed of data processing have become essential factors in healthcare decision-making. The demand for deeper integration of mathematical algorithms, such as artificial intelligence (AI) and in particular, machine learning (ML) in healthcare became evident. Already in 2020, 69% of medical institutions in the US and Europe started pilot AI projects. The boom in big data analysis tools and the availability of cheap computing power only strengthened this trend.
One specific application of AI in medicine is medical image analysis. Although a relatively narrow market at present, it addresses severe health conditions and diseases like cancer or COVID-19 complications. Moreover, AI/ML has the potential to revolutionize clinical studies by enabling researchers to extract insights from large and complex datasets more efficiently and accurately than traditional statistical methods.
Advances in computer vision technology have made imagery-based technology accurate and reliable enough for practical healthcare applications. It was already accepted as an essential tool in the healthcare industry to automate and improve treatments' success rates and empower medical experts to help more people. Medical imaging is also known as computer vision in medicine. Despite some conceptual differences between the terms “computer vision” and “image recognition” or “imaging,” for the purpose of this article, we treat both terms equally.
The global medical imaging market is expected to grow from $37.97B in 2021 to $56.53B in 2028, at a CAGR of 5.8%.
Medical imaging is a promising yet challenging market. Let's delve deeper into it and see why investors look at AI applications in medicine with growing interest.
One specific application of AI in medicine is medical image analysis. Although a relatively narrow market at present, it addresses severe health conditions and diseases like cancer or COVID-19 complications. Moreover, AI/ML has the potential to revolutionize clinical studies by enabling researchers to extract insights from large and complex datasets more efficiently and accurately than traditional statistical methods.
Advances in computer vision technology have made imagery-based technology accurate and reliable enough for practical healthcare applications. It was already accepted as an essential tool in the healthcare industry to automate and improve treatments' success rates and empower medical experts to help more people. Medical imaging is also known as computer vision in medicine. Despite some conceptual differences between the terms “computer vision” and “image recognition” or “imaging,” for the purpose of this article, we treat both terms equally.
The global medical imaging market is expected to grow from $37.97B in 2021 to $56.53B in 2028, at a CAGR of 5.8%.
Medical imaging is a promising yet challenging market. Let's delve deeper into it and see why investors look at AI applications in medicine with growing interest.
Benefits of AI in Medical Imaging
Medical imaging encompasses several non-invasive and invasive technologies that allow healthcare professionals to visualize different parts of a patient's body. The initial image data can be X-ray, ultrasound, MRI, CT, OCT scans, tactile imaging, endoscopy, and others. By incorporating AI, medical imaging provides insights into potential abnormalities, injuries, health issues, pathologies and pathological signs. It provides an in-depth visualization and allows medical professionals for more accurate diagnoses, improved decision-making, early detection, and health monitoring.
Value of AI implementation in medical imaging:
Value of AI implementation in medical imaging:
- Early detection. AI-based timely diagnosis and treatment before an issue develops to more advanced stages, based on early traces unnoticeable to the human eye. Medical imaging can detect various health conditions, including tumors, strokes, multiple sclerosis, fractures, pneumonia, and diagnose and stage various types of cancer. AI technologies help professionals identify minor abnormalities with a much more rapid turnaround time.
- Detecting rare pathologies. When a health state occurs only once in 100,000 cases, it may be overlooked even by experienced doctors. Due to their low frequency, rare health states may only sometimes be top of mind when providing diagnoses, and it's easier to recognize their unique traits with prior exposure during scholarship or study. However, AI accumulates vast knowledge and never «forgets» to consider even the rarest pathologies. It identifies subtle patterns that may be missed by human observation alone.
- Improved decision-making. Doctors need as much information as possible to provide an accurate diagnosis or procedure. Medical imaging has aided healthcare with more detailed imagery and reliable information, leading to more accurate diagnoses and decreased chances of misidentifying an issue. AI technologies may serve as a source of second «opinion» for the medical expert, supporting her/his own medical conclusion.
- Monitoring patients' state. Monitoring patients and analyzing their health condition statistics to decide if an operation is needed. Tracking post-iperational recovery over time and ensuring when the patients are fully recovered. Instructing them to follow a routine to ease their condition.
- Automation and turnaround. Medical imaging significantly increases medical professionals’ productivity. Their workload is reduced, as they can focus on parts of the scan highlighted by the AI and by this develop a faster diagnosis turnaround time.
Challenges of AI Implementation in medical imaging
While AI has significant potential in the healthcare industry, implementing it comes with various challenges.
Challenges in market and community strategy:
Challenges concerning data:
The more use cases we see on the market, the more optimistic we are about the market of AI in medical imaging.
Challenges in market and community strategy:
- Specialized AI solutions may have limited potential in the market. The available market for AI health apps is often small, and companies should research the market before launching their products.
- Companies need to focus on presales and implementing their products in healthcare organizations, rather than just improving the product.
- Vendors need to consider integrating their solutions into existing systems. Most healthcare management and diagnostic processes are already automated. If AI developers don’t think about how their solutions will integrate with these systems to get into clinics, they will never get there.
- Designing solutions that are friendly for medical staff. This requires more than just a user-friendly interface and easy onboarding. It’s also transparency and engagement in development and implementation. Medical staff must at least trust the technology and the system, but the best is to convert them to tech evangelists.
Challenges concerning data:
- Companies need to abide by data regulations and ensure data collection, access, and management standardization. Access to sensitive data should be secure, anonymous, and compliant with all relevant laws.
- Creating sufficiently large datasets is crucial. AI-based clinical practices require interconnected patient datasets to identify rare diseases accurately. However, obtaining centralized data for training can be challenging due to varying local data protection laws.
- Valid data labeling is essential and must be performed by skilled clinicians to ensure accuracy and validity. Inaccuracies can have fatal consequences for patients.
- Human physiology varies by geography. To use AI in a particular population, the model must be trained on a sample with this population adequately represented in it.
- Up-training is critical as medical professionals consistently provide AI with new data and improve diagnostic and therapeutic approaches. This makes medical imaging more efficient and accurate. Implementing tools for training is therefore essential.
The more use cases we see on the market, the more optimistic we are about the market of AI in medical imaging.
Market leaders
Major tech corporations are among those developing healthcare AI-powered solutions:
Several startups are also making significant strides in medical imaging:
Despite the challenges, the medical imaging market continues to grow, and we believe that investing intelligently in this market will yield benefits.
- Google develops solutions for imaging and diagnostic data, genomic research, nanoengineering, lung and breast cancer detection, and eye and skin disease treatment.
- Deepmind, a company acquired by Google, develops eye scan analysis for signs of blindness, head and neck cancer diagnostics with CT and MRI scans, breast cancer detection with mammography, and clinical mobile apps related to EMR to treat acute kidney injury.
- Siemens Healthineers used AI to detect signs of cancer from chest CT scans and played an important role during the COVID-19 pandemic.
- Major medical imaging equipment producers, like Carl Zeiss AG or Canon, also have their AI departments working on medical image analysis.
Several startups are also making significant strides in medical imaging:
- Paige (founded in 2017, raised $220M in VC). A pathologist analyses biopsy images and identifies suspicious areas for further study. The company has data from 1,000 institutions in 45 countries from 15,000 patients.
- Proscia (founded in 2014, raised $72.6M in VC). Analysis of tissue samples based on own and third-party software.
- Iterative Scopes (founded in 2017, raised $193.6M in VC). Gastroenterology, polyp detection.
- Shukun (founded in 2017). AI-powered disease diagnosis and treatment implemented in more than 400 hospitals in China. Datasets of over 100,000 cases.
- Lunit (founded in 2013, raised $133.5M in VC). Analysis of X-rays and detection of 10 major chest diseases, including cancer, with 97-99% accuracy.
- Aidoc (founded in 2016, raised $237.5M in VC). Analysis of medical imaging to detect whole-body organ abnormalities.
Despite the challenges, the medical imaging market continues to grow, and we believe that investing intelligently in this market will yield benefits.
Investing in AI in Medical Imaging
For our portfolio, we selected Altris AI. This is a unique AI-powered ophthalmic image management system. Altris AI is a co-pilot for medical professionals, allowing them to treat patients faster and more reliably. Altris AI improves the diagnostic process for ophthalmologists and optometrists by detecting more than 100+ retina pathologies and pathological signs, such as wet AMD or glaucoma in less than 60 seconds. With Altris AI, ophthalmologists, and optometrists perform OCT examinations faster, don’t miss minor, early, or rare pathologies, and have a second opinion when analyzing complex OCT scans.
The solution is fully GDPR compliant. Data is encrypted and can’t be reached by any 3rd party. They have a CE certificate and FDA clearance is in the process.
To ease the threshold of introducing tech solutions in non-digital native industries, like medicine, Altris AI offers free trials to allow medical institutions and eyecare businesses to experience the solution first-hand.
As the medical imaging industry matures and overcomes its primary complications, we believe it will grow fast and keep a close eye on its development.
The solution is fully GDPR compliant. Data is encrypted and can’t be reached by any 3rd party. They have a CE certificate and FDA clearance is in the process.
To ease the threshold of introducing tech solutions in non-digital native industries, like medicine, Altris AI offers free trials to allow medical institutions and eyecare businesses to experience the solution first-hand.
As the medical imaging industry matures and overcomes its primary complications, we believe it will grow fast and keep a close eye on its development.
Key takeaways
* During the COVID pandemic, the shortage of expert doctors triggered AI-based digitalization in medicine that would enable doctors to increase their turnaround.
* One area where AI is already showing promise in medicine is image recognition. It’s used for early detection, detecting rare pathologies, augmented decision-making, monitoring, and automation.
* Challenges to overcome are developing effective go-to-market strategies, complying with data regulations, and providing valid data.
* Market leaders in medical imaging include startups and dedicated departments or subdivisions of major tech companies such as Google and Siemens.
* To explore the potential of the medical imaging market, we invested in Altris AI, an AI-powered system that performs ophthalmic image or retina scan analysis and assists ophthalmologists and optometrists in diagnostic decision-making.
* One area where AI is already showing promise in medicine is image recognition. It’s used for early detection, detecting rare pathologies, augmented decision-making, monitoring, and automation.
* Challenges to overcome are developing effective go-to-market strategies, complying with data regulations, and providing valid data.
* Market leaders in medical imaging include startups and dedicated departments or subdivisions of major tech companies such as Google and Siemens.
* To explore the potential of the medical imaging market, we invested in Altris AI, an AI-powered system that performs ophthalmic image or retina scan analysis and assists ophthalmologists and optometrists in diagnostic decision-making.
