Next week, PathAI scientists Sean Grullon and Marc Thibault will be presenting at four workshops at [ICML] Int'l Conference on Machine Learning July 21-27 in Vienna, Austria! View the program: https://lnkd.in/eTNqB8i6 If you’re attending the conference, stop by the workshops and hear how #MachineLearning was used to develop the new #pathology-centric #FoundationModel PLUTO for quantitative histopathology at scale and learn more about how it was used to reveal deeper biological insights. Read the abstracts and add the workshops to your schedule below. 1. #PLUTO: Pathology Universal Transformer Pathology images provide a unique challenge for computer-vision-based analysis: a single whole slide image is gigapixel-sized and often contains hundreds of thousands to millions of objects of interest across multiple resolutions. In this work, we propose PathoLogy Universal TransfOrmer (PLUTO): a light-weight pathology foundation model (FM) that is pre-trained on a diverse dataset of 195 million image tiles collected from multiple sites. We design task-specific adaptation heads that utilize PLUTO's output embeddings for tasks ranging from subcellular- to slide-scale, including instance segmentation, tile classification, and slide-level prediction. We find that PLUTO matches or outperforms existing task-specific baselines and pathology-specific FMs, some of which use orders-of-magnitude larger datasets and model sizes. Our findings present a path towards a universal embedding to power pathology image analysis, and motivate further exploration around pathology FMs in terms of data diversity, architectural improvements, sample efficiency, and practical deployability in real-world applications. Workshops: - Accessible and Efficient Foundation Models for Biological Discovery - ICML 2024 Workshop on Foundation Models in the Wild - Machine Learning for Life and Material Science: From Theory to Industry Applications - Publication: https://lnkd.in/ewkaGj3D 2. Interpretability analysis on a pathology foundation model reveals biologically relevant embeddings across modalities Mechanistic interpretability has been explored in detail for large language models (LLMs). For the first time, we provide a preliminary investigation with similar interpretability methods for medical imaging. Specifically, we analyze the features from a ViT-Small encoder obtained from a pathology Foundation Model via application to two datasets: one dataset of pathology images, and one dataset of pathology images paired with spatial transcriptomics. We discover an interpretable representation of cell and tissue morphology, staining patterns, and gene expression within the model embedding space. Our work paves the way for further exploration around interpretable feature dimensions and their utility for medical and clinical applications. Workshop: ICML 2024 Workshop on Mechanistic Interpretability Publication: https://lnkd.in/eD8JhrFs #ICML #ICML2024
PathAI
Software Development
Boston, Massachusetts 52,511 followers
Improving patient outcomes with AI-powered pathology.
About us
PathAI's mission is to improve patient outcomes with AI-powered pathology. Our platform promises substantial improvements to the accuracy of diagnosis and the efficacy of treatment of diseases like cancer, leveraging modern approaches in machine learning. We are a company of diverse employees with a wide range of backgrounds and experiences. Our world class team is passionate about solving challenging problems and making a huge impact. Our office is located in the heart of Fenway. PathAI was recently voted one of BBJ's Best Places to Work!
- Website
-
http://www.pathai.com
External link for PathAI
- Industry
- Software Development
- Company size
- 501-1,000 employees
- Headquarters
- Boston, Massachusetts
- Type
- Privately Held
- Founded
- 2016
- Specialties
- artificial intelligence, pathology, digital pathology, oncology, immuno-oncology, auto immune disease, neurodegenerative disease, companion diagnostics, precision medicine, solutions, R&D, computational pathology, deep learning, software engineering, AI, Machine Learning, IBD, and cancer research
Locations
-
Primary
Get directions
1325 Boylston St
Suite 10000
Boston, Massachusetts 02215, US
-
Get directions
3495 Hacks Cross Rd
Memphis, Tennessee 38125, US
Employees at PathAI
Updates
-
PathAI reposted this
Yesterday, a team of PathAI scientists led by Nhat Le, John Abel, Sean Grullon, and Dinkar Juyal published "𝐈𝐧𝐭𝐞𝐫𝐩𝐫𝐞𝐭𝐚𝐛𝐢𝐥𝐢𝐭𝐲 𝐚𝐧𝐚𝐥𝐲𝐬𝐢𝐬 𝐨𝐧 𝐚 𝐩𝐚𝐭𝐡𝐨𝐥𝐨𝐠𝐲 𝐟𝐨𝐮𝐧𝐝𝐚𝐭𝐢𝐨𝐧 𝐦𝐨𝐝𝐞𝐥 𝐫𝐞𝐯𝐞𝐚𝐥𝐬 𝐛𝐢𝐨𝐥𝐨𝐠𝐢𝐜𝐚𝐥𝐥𝐲 𝐫𝐞𝐥𝐞𝐯𝐚𝐧𝐭 𝐞𝐦𝐛𝐞𝐝𝐝𝐢𝐧𝐠𝐬 𝐚𝐜𝐫𝐨𝐬𝐬 𝐦𝐨𝐝𝐚𝐥𝐢𝐭𝐢𝐞𝐬." This preprint is a continuation of our initiative to construct pathology-centric #foundationmodels and links PLUTO, PathAI’s foundation model, with interpretable aspects of tumor biology. https://lnkd.in/eD8JhrFs Foundation models are gaining traction in #pathology, however, the applicability of foundation models to diverse use cases rests on their ability to capture latent biology without supervised training. 𝐓𝐨 𝐭𝐞𝐬𝐭 𝐭𝐡𝐢𝐬, 𝐰𝐞 𝐟𝐨𝐜𝐮𝐬𝐞𝐝 𝐨𝐧 𝐭𝐡𝐫𝐞𝐞 𝐚𝐩𝐩𝐫𝐨𝐚𝐜𝐡𝐞𝐬 𝐟𝐨𝐫 𝐥𝐢𝐧𝐤𝐢𝐧𝐠 𝐏𝐋𝐔𝐓𝐎 𝐰𝐢𝐭𝐡 #biology: 1. Relating PLUTO embeddings to spatial #transcriptomics across multiple cancer types. 2. Relating PLUTO embeddings to cell type and morphology. 3. Interrogating PLUTO’s embeddings with a sparse autoencoder–revealing that PLUTO embeddings encode interpretable aspects of the WSI beyond what is typically analyzed in digital pathology. Most excitingly (to us!) we found that deconstructing PLUTO embeddings with the sparse autoencoder revealed unexpected structure in the embeddings. This structure captured subtle aspects of patches from whole slide images (WSIs) including cell morphological subtypes, tissue geometry and collagen alignment, and even tissue preparation characteristics such as small amounts of surgical ink–shown in the figure below. Taken together, these results ground PLUTO in fundamental tumor biology and improve our confidence that PLUTO’s embeddings are biologically interpretable, powerful, and general for downstream tech applications. You can read the preprint here: https://lnkd.in/eD8JhrFs #CancerResearch #Biotech #MachineLearning #DeepLearning #AI #FoundationModels #Pathology #SpatialBiology
-
Yesterday, a team of PathAI scientists led by Nhat Le, John Abel, Sean Grullon, and Dinkar Juyal published "𝐈𝐧𝐭𝐞𝐫𝐩𝐫𝐞𝐭𝐚𝐛𝐢𝐥𝐢𝐭𝐲 𝐚𝐧𝐚𝐥𝐲𝐬𝐢𝐬 𝐨𝐧 𝐚 𝐩𝐚𝐭𝐡𝐨𝐥𝐨𝐠𝐲 𝐟𝐨𝐮𝐧𝐝𝐚𝐭𝐢𝐨𝐧 𝐦𝐨𝐝𝐞𝐥 𝐫𝐞𝐯𝐞𝐚𝐥𝐬 𝐛𝐢𝐨𝐥𝐨𝐠𝐢𝐜𝐚𝐥𝐥𝐲 𝐫𝐞𝐥𝐞𝐯𝐚𝐧𝐭 𝐞𝐦𝐛𝐞𝐝𝐝𝐢𝐧𝐠𝐬 𝐚𝐜𝐫𝐨𝐬𝐬 𝐦𝐨𝐝𝐚𝐥𝐢𝐭𝐢𝐞𝐬." This preprint is a continuation of our initiative to construct pathology-centric #foundationmodels and links PLUTO, PathAI’s foundation model, with interpretable aspects of tumor biology. https://lnkd.in/eD8JhrFs Foundation models are gaining traction in #pathology, however, the applicability of foundation models to diverse use cases rests on their ability to capture latent biology without supervised training. 𝐓𝐨 𝐭𝐞𝐬𝐭 𝐭𝐡𝐢𝐬, 𝐰𝐞 𝐟𝐨𝐜𝐮𝐬𝐞𝐝 𝐨𝐧 𝐭𝐡𝐫𝐞𝐞 𝐚𝐩𝐩𝐫𝐨𝐚𝐜𝐡𝐞𝐬 𝐟𝐨𝐫 𝐥𝐢𝐧𝐤𝐢𝐧𝐠 𝐏𝐋𝐔𝐓𝐎 𝐰𝐢𝐭𝐡 #biology: 1. Relating PLUTO embeddings to spatial #transcriptomics across multiple cancer types. 2. Relating PLUTO embeddings to cell type and morphology. 3. Interrogating PLUTO’s embeddings with a sparse autoencoder–revealing that PLUTO embeddings encode interpretable aspects of the WSI beyond what is typically analyzed in digital pathology. Most excitingly (to us!) we found that deconstructing PLUTO embeddings with the sparse autoencoder revealed unexpected structure in the embeddings. This structure captured subtle aspects of patches from whole slide images (WSIs) including cell morphological subtypes, tissue geometry and collagen alignment, and even tissue preparation characteristics such as small amounts of surgical ink–shown in the figure below. Taken together, these results ground PLUTO in fundamental tumor biology and improve our confidence that PLUTO’s embeddings are biologically interpretable, powerful, and general for downstream tech applications. You can read the preprint here: https://lnkd.in/eD8JhrFs #CancerResearch #Biotech #MachineLearning #DeepLearning #AI #FoundationModels #Pathology #SpatialBiology
-
Our Chief Medical Officer, Eric Walk MD, FCAP, presented on behalf of PathAI at the #AWSSummit this week in NYC. Dr. Walk shared insights into how PathAI is leveraging cloud infrastructure to unlock the next wave of #precisionmedicine innovation. Using examples from our cutting-edge AISight® image management system and platform, he highlighted the transformative power of #digitalpathology and #AI. At PathAI, we believe the future of patient care will be revolutionized by digital pathology and artificial intelligence, driving better patient outcomes, enhanced diagnostic accuracy, and operational efficiency. Check out some highlights from his presentation and see how PathAI is leading the way in AI-driven and digital diagnostics. Link https://lnkd.in/eR8dWEWu #PathAI #AWSsummit #PrecisionMedicine #DigitalPathology #AI #HealthcareInnovation #Diagnostics #Healthcare #Pathology AISight® is for research use only. Not for use in diagnostic procedures.
-
-
🔎 Interested in analyzing the tissue microenvironment in inflammatory bowel disease? View #IBD #pathology samples through the enhanced lens of #AI with IBD Explore™! You can now access our new IBD Explore™ demo to see how #AI-powered digital pathology can supercharge your #digestivedisease research. IBD Explore™ enables a detailed quantification of tissue regions, cellular composition, and inflammatory infiltration in H&E-stained #UC and #CD biopsies. Human-interpretable features from IBD Explore™ can be used to explore tissue composition, quantify treatment effect and mechanism of action, or identify possible #biomarkers predictive of drug response. Sign up for an account to access our demo; if you've already registered an account, the new slides are already available! https://lnkd.in/eCn4xwj2 #pathology #IBD #AI #digitalpathology #precisionmedicine IBD Explore™ is for research use only. Not for use in diagnostic procedures.
-
🗞 We're thrilled to announce the launch of our new quarterly newsletter, designed to keep you at the forefront of AI-powered #pathology advancements. Stay up to date with the latest PathAI #digitalpathology innovations, announcements, publications, and more. Whether you're a pathologist, researcher, or simply passionate about the future of #medicine, our newsletter is your gateway to understanding how AI is revolutionizing pathology. Sign up: https://lnkd.in/ezFEqmeK #AI #biotech #MachineLearning #DeepLearning #Oncology #biotechnology
-
PathAI reposted this
Great writeup about our new paper from PathAI -- if you're interested in nuclear morphology and want to see it in action (beyond the paper figures), it's available to interact with via our demos on some example TCGA slides https://lnkd.in/ecetsUqr
Freelance Medical & Scientific Writer | Oncology, Immunotherapy, mRNA Therapeutics, Targeted Therapies | Genomics, Molecular Diagnostics, Biomedical Data Science | Biotech Product Marketing, Publications, Medical Affairs
Alterations in nucleus size, shape, and color are ubiquitous in cancer, but comprehensive quantification of nuclear morphology across a whole-slide histologic image remains challenging. A new open-access publication by John Abel and larger team at PathAI details the development of a pan-tissue, deep learning-based digital pathology pipeline for exhaustive nucleus detection, segmentation, and classification. I included the link to the full paper and a summary below for anyone interested. AI powered quantification of nuclear morphology in cancers enables prediction of genome instability and prognosis. https://lnkd.in/esNTTDjs Methods overview: The authors collected over 29,000 manual annotations of cell nuclei from H&E images of 21 tumor types and other diseases to train and validate an object detection and segmentation model. Their model for nucleus analysis is a Mask-RCNN-style architecture with a ResNet50 backbone pretrained on ImageNet. It was deployed on whole-slide H&E images from breast cancer (TCGA BRCA), lung adenocarcinoma (TCGA LUAD), and prostate adenocarcinoma (TCGA PRAD) cohorts. Following nuclear segmentation, the cell class of each nucleus was assigned using PathExplore models specific to each cancer type. Interpretable features describing the size, shape, stain intensity, and texture were computed for each individual nucleus. The mean and standard deviation of each feature for each cell class on each slide were used to summarize the nuclear morphology. Statistical analyses were performed to assess relationships between nuclear morphology and cancer type, genomic instability, breast cancer molecular subtype, survival, and gene expression. Results overview: The nuclear segmentation and classification model performed comparably to previously reported models. It revealed differences in nuclear morphology sufficient to distinguish between BRCA, LUAD, and PRAD. Cancer cell nuclear area was associated with increased aneuploidy score and homologous recombination deficiency across cancer types. It was also predictive of whole genome doubling. In the breast BRCA cohort, cell-type-specific nuclear morphology enabled classification of luminal A, basal-like, and HER2 molecular subtypes. Increased fibroblast nuclear area in BRCA was indicative of poor progression-free and overall survival and was associated with gene expression signatures related to extracellular matrix remodeling and anti-tumor immunity. In summary, this work highlights the power of machine learning-driven quantitative nuclear morphometry in multiple cancer types at scale. The models and resulting features have the potential to aid pathologists in discerning novel biomarkers and provide meaningful prognostic information for cancer patients.
-
-
🎆 Happy Independence Day from all of us at PathAI! 🎆 As we celebrate the 4th of July, we reflect on the freedoms we enjoy individually and as a collective, and how they enable the work we do. Thank you for partnering with us to innovate #pathology.
-
🔥 Kick off summer with a Fireside Chat on #DigitalPathology featuring our CEO, Andrew Beck! Join Andy and an incredible lineup of #GoogleCloud experts, partners, and customers at the Healthcare and Life Sciences Summer Camp. They'll dive into the latest #HealthTech #AI trends. https://lnkd.in/ecV5VK_R Sign up today to hear Andy’s discussion on July 24th at 1:00 PM ET to discover how PathAI is leveraging cutting-edge #AI technology to revolutionize #Pathology, #DrugDevelopment, #Oncology, #IBD, and #LiverDisease research.
-
This past week, we published our article, 𝐀𝐈 𝐩𝐨𝐰𝐞𝐫𝐞𝐝 𝐪𝐮𝐚𝐧𝐭𝐢𝐟𝐢𝐜𝐚𝐭𝐢𝐨𝐧 𝐨𝐟 𝐧𝐮𝐜𝐥𝐞𝐚𝐫 𝐦𝐨𝐫𝐩𝐡𝐨𝐥𝐨𝐠𝐲 𝐢𝐧 𝐜𝐚𝐧𝐜𝐞𝐫𝐬 𝐞𝐧𝐚𝐛𝐥𝐞𝐬 𝐩𝐫𝐞𝐝𝐢𝐜𝐭𝐢𝐨𝐧 𝐨𝐟 𝐠𝐞𝐧𝐨𝐦𝐞 𝐢𝐧𝐬𝐭𝐚𝐛𝐢𝐥𝐢𝐭𝐲 𝐚𝐧𝐝 𝐩𝐫𝐨𝐠𝐧𝐨𝐬𝐢𝐬, in npj Precision Oncology where we describe the development of of a pan-tissue, #deeplearning based digital pathology pipeline for nucleus detection, segmentation, and classification enhancing nuclear morphologic #biomarker discovery. https://lnkd.in/eJwF_YZH Did you know that quantification of nuclear morphology is available as part of the PathExplore™ suite of algorithms? Here it is in action! Dive deep into nuclear morphology, cell & tissue segmentation, and more with PathExplore™ for #cancerresearch. Access our demo slides on our AISight® TxR platform so you can see all of the capabilities of AI-powered pathology has to offer: https://lnkd.in/eCn4xwj2 PathExplore™ is for research use only. Not for use in diagnostic procedures. #pathology #biotech #AI #biotechnology #oncology