Biomedical imaging and artificial intelligence (2nd part)
Biomedical imaging and artificial intelligence (2nd part)
Applying Deep Learning to Biomedical Imaging
For the first part of this newsletter:
Key players in AI-assisted Imaging
According to a GlobalData’s analysis, there are 80 companies, spanning technology vendors, established technology companies and up-and-coming start-ups engaged in the development and application of AI-assisted microscopy. For example, Heartflow is a leading patent filer in AI-assisted microscopy, for systems determining cardiovascular information of a patient. The other prominent top 10 patent filers in the space include F. Hoffmann-La Roche, Siemens, Olympus, Fujifilm, Koninklijke Philips, Danaher, Hologic, Sony Group and Sysmex. In terms of geographical reach, Terumo leads the pack, followed by Room4 Group and ClearLight Biotechnologies. And in terms of application diversity, Alentic Microscience holds the top position, followed by Horiba and CellPly.
According to Lumea—offering a state-of-the-art Image Management System with industry-leading AI tools from Verily, an Alphabet precision health company, designed to empower diverse histopathology applications—the top 10 companies that have integrated AI in Digital Pathology are: Aiforia, AIRA Matrix in India, Deep Bio in South Korea, DoMore Diagnostics in Norway, Indica Labs in UK, Mindpeak in Germany, Paige AI in US, PathAI in US, Tempus in US and Verily an Alphabet company.
The top emerging Medical Imaging startups according StartUs Insights—that empowers you to access the world's information on innovation, emerging companies, and technologies—are:
the Brazilian startup Medical Harbour that develops medical imaging technologies and solutions for radiology and teleradiology (Athena Hub),
the Taiwanese startup AetherAI that combines digital microscopy and AI to offer insights on medical images (making Whole-Slide Image Solutions standard and using AI-Microscope to conduct medical examinations),
the Zimbabwe-based startup Dr CADx that develops a DL-based application to interpret medical images with high accuracy, and
the US-based startup Exo Imaging that develops an AI-based platform for ultrasound image analysis (Exo Iris delivers high-performance imaging that fits in your pocket).
According to AIMultiple—featuring AI solutions in analytics, marketing, sales, customer service, tech, IT, fintech, healthcare, retail, e-commerce and other sectors—the top companies in AI-powered medical imaging for 2024 are:
Butterfly Network: offers Butterfly IQ, a portable mobile device that uses ultrasound-on-chip technology which makes it the world’s first handheld entire body ultrasound framework, that they claim has the capability of detecting diseases in real-time while scanning.
Arterys, now part of Tempus: built the first tech product to visualize & quantify blood flow in the body using any MRI, and is a pioneer in 4D cloud-based imaging. Arterys’ Lung-AI platform helps to reduce missed detections by 42 to 70%.
Gauss Surgical part of Stryker: received CE Mark for its Triton System for iPad, the world’s first and only mobile platform for real-time monitoring of surgical blood loss.
Sigtuple in India: uses robotics and AI to digitize any biological sample on a glass slide to enable AI aided remote review.
Freenome: works on detecting cancer by imaging blood cells.
Enlitic: uses DL techniques to analyze the data extracted from radiology images.
Caption Health: provides guidance to healthcare professionals and inexperienced people to perform ultrasound examinations accurately and quickly.
Behold.ai: uses AI to help radiologists diagnose radiology scans in a variety of cases.
Viz.ai: sends a notification to healthcare professionals when there is a sign of a serious situation, like early signs of brain stroke.
RetinAi‘s: has a discovery platform that helps to collect, organize and analyze health data from the eye.
Subtle Medical: improves the quality of noisy medical images and provides better interpretation.
BrainMiner: is a UK based company and its software DIADEM provides an automated system for analyzing MR brain scans. And
Lunit: has developed AI solutions for precision diagnostics and therapeutics.
Image resolution and AI
In general, the AI models in medical imaging can be used for data augmentation, image enhancement, image reconstruction & segmentation, and anomaly detection in areas such radiology, pathology, surgical planning and disease progression modeling. Image resolution—the level of detail contained in an image, that more specifically it refers to the number of pixels that exist within an image, and the higher the resolution and the richer the pixel count, the more detail and definition you will see—is another image characteristic where AI image upscalers can use the power of AI to preserve or even enhance the resolution of the photos.
In microscopy, the term “resolution” is used to describe the ability of a microscope to distinguish the minimum distance between 2 distinct points of a specimen where they can still be seen by the observer or microscope camera as separate entities.
Improving the spatial resolution of a fluorescence microscope has always been a challenge in the imaging community. To address this issue, a variety of approaches have been taken, like for example deconvolution, where images are numerically deblurred based on a knowledge of the microscope point spread function. However, deconvolution can easily lead to noise-amplification artifacts. Also deblurring by post-processing can lead to negativities or fail to conserve local linearity between sample and image.
Accordingly, in this paper Resolution enhancement with deblurring by pixel reassignment (DPR) two scientists (Bingying Zhao and Jerome Mertz) just described a simple image deblurring algorithm based on pixel reassignment that inherently avoids artifacts and can be applied to general microscope modalities and fluorophore types, allowing to distinguish nearby fluorophores even when these are separated by distances smaller than the conventional resolution limit, helping facilitate for example, the application of single-molecule localization microscopy in dense samples.
AI/ML solutions and tools for imaging
⚙ Onc.AI (US, 2020) is a privately held digital health company on a mission to radically improve oncology decision making. By leveraging a market-leading oncology real-world dataset (diagnostic imaging, EMR, labs, genomics), Onc.AI is developing a pipeline of AI models to fulfill the potential of precision oncology. Their first product will help medical oncologists make treatment decisions for metastatic lung cancer patients treated with PD-1 immunotherapy. Onc.AI has raised a total funding of $31M over 2 rounds from 13 investors.
⚙ Brainomix, a spin-out from the University of Oxford in 2010, is an AI-powered MedTech solutions company that has pioneered the development of an AI platform that automates validated imaging biomarkers to improve both diagnosis and treatment decisions. For example, the e-Stroke platform (e-Aspects, e-CTA and e-CTP) is the world's most comprehensive stroke imaging solution, and the e-ILD technology is an automated AI software which has been trained to process high resolution chest CT data in patients with interstitial lung diseases, that cause progressive pulmonary fibrosis.
On March 21, 2023, Brainomix announced that its 360 e-ASPECTS tool for assessing stroke signs on plain CT brain scans has received FDA clearance, enabling the company to deploy its cutting-edge AI imaging platform to stroke centers in the USA. Brainomix is also offering a Cancer solution that has been trained to characterize and quantify solid-tumors using thousands of cases from leading academic institutions, unlocking objective measurements of tumor lengths and other biomarkers such as tumor volume.
Brainomix has also a collaboration with the French medical neurology imaging company Pixyl for distributing Pixyl’s SaaS solution Neuro.MS for the diagnosis and treatment of multiple sclerosis (MS). In particular, Pixyl.Neuro is a cutting-edge medical technology that harnesses the power of advanced AI to detect anomalies and changes in brain MR images for neuroinflammatory and neurodegenerative disorders:
Pixyl.Neuro.MS automatically analyzes MRI images and provides structured report on multiple sclerosis, and
Pixyl.Neuro.BV provides detailed information on brain anatomy quantification used for many pathologies, including: dementias, psychiatric disorders and multiple sclerosis.
Moreover, Brainomix and Bridge Biotherapeutics—a clinical-stage biotechnology company focused on developing novel drugs for cancer, fibrosis and inflammation—announced a new partnership to deliver a quantitative imaging biomarker analysis within a phase 2 study. Under this partnership, Brainomix will leverage its e-ILD technology—automated AI software—which has been trained to process high resolution chest CT data in patients with interstitial lung diseases that cause progressive pulmonary fibrosis.
Brainomix has raised a total funding of $34.8M over 5 rounds from 6 investors.