Welcome back to another edition of MetaphysicalCells. Today is all about AI being used to mine visual information and reveal patterns that the human eye cannot even grasp.
AI-assisted bioimaging news 📰
📌 Avi Veidman the CEO of Nucleai, an AI-powered image analysis company in Israel, just announced Nucleai’s partnership 🤝 with Mayo Clinic to advance spatial biology in biopharma. By integrating Mayo's rich datasets, laboratories and expertise with Nucleai's cutting-edge AI platform the two partners aim to revolutionise the precision medicine landscape and transform digital pathology.
Nucleai that has access to millions of annotated cells across more than ten types of cancers, through its algorithms can detect and characterise all kind of tissues and cell architectures. Afterwards deep-learning models combining spatial features and clinical data can drive the discovery of pathology based spatial biomarkers. In particular, the Nucleai Atom platform transforms pathology slides into raw data of cells and area locations, and then the raw data is fed into the Nucleai engine computing hundreds of spatial features intersected with outcome to predict drug response.
Nucleai’s AI-powered digital assays are mainly used to incorporate qualified biomarkers into prospective clinical trials or are used as companion diagnostics across the world in partnership with biopharma, diagnostics and clinical laboratories.
Last week, Nucleai also announced a strategic partnership with the biopharmaceutical firm Adlai Nortye (a global clinical-stage biopharmaceutical company focused on innovative oncology drugs) aimed at identifying, validating and testing novel hematoxylin and eosin staining-based biomarkers in Adlai Nortye’s clinical trials.
📌 The Van Valen Lab at the California Institute of Technology (Caltech) has created a suite of AI models, available at deepcell.org, to count and analyse cells and other features from images both of live cells and preserved tissues. Deepcell’s platform combines AI, proprietary microfluidics, high resolution optics and a growing cell atlas with over 1 billion images to analyse and sort cells based on visual features.
On May 17, 2023, Deepcell 📢 announced the launch of the REM-I Platform, a cell morphology analysis and sorting platform which comprises the REM-I benchtop instrument with Human Foundation Model and Axon data suite, bringing together single cell imaging, sorting and high-dimensional analysis (Deepcell Launches AI-Powered Single Cell Analysis Platform to Accelerate Cell Biology Discovery and Catalyze Field of Morpholomics).
In particular, the Human Foundation Model (HFM) characterises each cell in real-time by combining self-supervised learning and morphometrics (computer vision) to extract a 115-dimensional embedding vector representing cell morphology. Then the embeddings representing each cell’s unique morphology are projected onto a morphology UMAP (or Uniform Manifold Approximation and Projection, that is a dimension reduction technique that can be used for visualisation) in real-time, and clustered by morphology so that in the Axon data suite users can view, compare and sort morphologically distinct cell populations.
Interestingly, the REM-I platform is offering high-resolution images of single cells completely label-free (that is increasing cell viability and keeping the cells unperturbed) that are captured at high speed, while the AI characterisation of captured cell images happens in real-time.
📌 Brainomix, a spin-out from the University of Oxford, 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 centres in the USA (Brainomix receives US FDA clearance for flagship stroke AI imaging software).
Brainomix is offering also a Cancer solution that has been trained to characterise and quantify solid-tumours using thousands of cases from leading academic institutions, unlocking objective measurements of tumour lengths and other biomarkers such as tumour volume.
Finally, 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 analyses MRI images and provides structured report on MS and Pixyl.Neuro.BV provides detailed information on brain anatomy quantification used for many pathologies, including: dementias, psychiatric disorders and multiple sclerosis.
📌 Digital Diagnostics (US) has an AI tool for detecting early signs of diabetic retinopathy in photos of the eyes 👀 that are captured using retinal cameras. Previously known as IDx-DR, the LumineticsCore is an FDA cleared autonomous AI diagnostic system that has been installed in nine Labcorp patient service centres across Alabama to treat diabetic people (Digital Diagnostics launches LumineticsCore system in southeast US). LumineticsCore was the first FDA-cleared AI diagnostic system to make a diagnosis without physician input at the point-of-care.
📌 GE Healthcare has a new deep learning software that is engineered to provide increased small, low-contrast lesion detectability compared to the conventional Time-of-Flight PET/CT scanner. Precision DL, that uses the power of deep learning image processing to give image quality performance benefits, is using a convolutional neural network, which is designed to analyse large quantities of PET/CT images 🩻 and learn the characteristics of each one. Obtaining in this way high-sensitivity and high-resolution images, with up to 28% improvement in contrast recovery, up to 64% improvement in contrast to noise ratio and 14% improvement in feature quantitation accuracy.
On May 23, 2023, the GE Healthcare announced US FDA 510(k) clearance of Precision DL (GE HealthCare Receives FDA Clearance of a New Deep Learning Solution for Enhanced Image Quality in PET/CT, Advancing its Leadership Position in AI) marking a new era for PET/CT performance and outcomes, enabling clinicians to decode coincidence events at exceptionally fine resolutions for informed diagnoses and treatment planning.
📌 Heartflow in California provides unprecedented visualisation of each patient’s coronary arteries, providing a 3D personalised model and analysis, all assisted by AI powered algorithms. Founded in 2010, their HeartFlow FFRCTAnalysis 🫀 tool for diagnosis and management of coronary artery disease is now commercially available in Europe, US, Canada and Japan. HeartFlow FFRCT is an AI-powered software that maps out the heart’s coronary arteries and any potential blockages through a 3D CT scan.
HeartFlow has just closed $215 million in new venture capital funding led by Bain Capital Life Sciences (Money continues to flow for HeartFlow with $215M VC raise), for a total raised of $792.7M.
📌 Celly (US) is an iphone-based AI microscope 🔬 app allowing to scan slides. Once the Celly app is downloaded on your iphone, you can use a LabCam Ultra adapter ($299) to connect your iphone to the microscope (you will need also a motorised stage $1400 $599) and then use the app to start scanning slides. At this point Celly instantly counts cells and detects abnormalities, while the supported tests are for Hematology (automated WBC, PLT and RBC count for blood and bone marrow smears), Malaria, PAP Smear (in beta) and Veterinary.
Before you use Celly you can choose between the following plans:
Free (15 tests per day, 10 Gb Storage, Telepathology),
Standard (80 tests per day, 1 Tb Storage and LIMS integration) at $199 per month, and
Premium (For unlimited tests, Flexible Storage and 24/7 Support) ask for a quote.
The AI Software for Microscopy Market is anticipated to expand at a CAGR of +9% by 2029 (report)
📌 ClearLight Biotechnologies (US) is offering an interesting solution to an almost 100 years old problem affecting histology. Current gold standard histological techniques, that are 100 years old, rely on the analysis of “Formalin-Fixed Paraffin-Embedded” (FFPE) 2D thin sections (~4-10 micron), that unfortunately translates to several problems:
the spatial and morphological analysis of the tissue microenvironment is highly limited,
the 2D analysis is destructive only slide-based, non-digital, variable and highly manual that does not represent the entire biology,
and finally if we want to be really honest there are also many technical challenges during the FFPE process.
For now, a solution to all these problems comes from the CLARITY technology from ClearLight that brings 3D histopathology closer to everyday research. CLARITY is an acronym for Clear Lipid-exchanged Acrylamide-hybridized Rigid Imaging / Immunostaining / in situ-hybridization-compatible Tissue hYdrogel) that embeds tissues in a hydrogel matrix, after lipids were cleared to create an optically transparent hybrid that is transparent to light and permeable to molecules. At this point, scientists can add markers for staining and further imaging, and finally the tissue processing workflow can also include washing out the initial staining allowing for multiple rounds of labelling and imaging.
Specifically, once the tissue is stained is placed in a mounting solution for imaging with a confocal or light sheet microscope or another 3D technique in order to overcome the limitations of the 2D imaging, which is not representative of the heterogeneous biopsy or the tissue microenvironment, while at the same time AI algorithms are trained with the 3D data.
ClearLight Biotechnologies was founded in 2015 and is offering a very interesting solution, that I would definitely try, and I am wondering what it will take (an asteroid—or perhaps a comet to end the age of the FFPE-dinosaurs 🦖) to change the centenary and old-fashioned FFPE process, if CLARITY ends up to be a better alternative.
Another innovative solution similar to CLARITY comes from a group of researchers from the University of Eastern Finland, the University of Turku, and Tampere University that have developed an AI virtual staining in order to replace the old chemical staining that has been the cornerstone of studying histopathology for more than a century (Researchers Developed an AI-Based Method to Replace Chemical Staining of Tissue). The AI method developed in this study produces computational images that very closely resemble those produced by the actual chemical staining process, and requires no special hardware or infrastructure beyond a regular light microscopy and a suitable computer.
📌 WEBKNOSSOS (the result of a collaboration between neuroscientists at the Max Planck Institute for Brain 🧠 Research in Germany) is an open-source tool for annotating and exploring large 3D image datasets. For example, neuroscientists can visualise and analyse neural tissue by uploading their data, access and annotate it from anywhere, and create segmentations collaboratively or reconstruct the brain's cells and map their connections. Or cell biologists can visualise their biological data, segment and analyse cells, to create segmentations collaboratively using volume and skeleton annotation tools or visualise their ML predictions.
The core of WEBKNOSSOS, for things like fast browsing of large 3D images, volume & skeleton annotation and data sharing, is free. For advanced features and tighter collaboration, such as visualisation of 3D meshes, dataset management, collaborative annotation & project management and segmentation proofreading, you can upgrade to Team or Power plans, from 5,000€ per year. Follow them on Twitter (@webknossos) as they constantly release new AI-based tools for imaging.
Artificial intelligence innovation: Leading companies in AI-assisted microscopy (Source)
📌 Finally, CellPly from Italy has VivaCyteTM a revolutionary fully automated analytical platform for profiling the immune cell function. Through the use of proprietary microfluidic technologies, highly-automated analytical instruments and AI-powered image and data analytics, the Cellply platform defines the function of the immune system one cell at a time, combining functional data with other multiplex data and providing an unprecedented single-cell analytical solution. The AI-assisted cell detection and classification, is followed by powerful dashboard enable deep single-cell multiplex and time-lapse analytics.
Until next time 🍨,
PS: For more AI biotech news from Substack:
By Decoding Bio