Gradient Health’s Post

Another day, another step forward. On March 5th, INESC TEC and IMP Diagnostics Molecular & Anatomic Pathology published the work behind the first prototype that uses #AI for colorectal diagnosis. 𝗪𝗵𝗮𝘁 𝗱𝗼𝗲𝘀 𝗶𝘁 𝗰𝗼𝗻𝘀𝗶𝘀𝘁 𝗼𝗳? 👉 This work focuses on improving a prototype that uses AI as a complementary tool to the diagnosis of colon and rectal biopsies, and the availability of the largest database of digital images of colorectal pathologies. The advances drive image analysis technology and contribute to the development of more effective solutions in the diagnosis of #ColorectalCancer. The researchers trained a new model using close to 10000 images of tissues with colorectal pathology, thus achieving a diagnostic acuity of 93.44% and a sensitivity of 99.7% in the detection of high-risk lesions related to this type of cancer. 📄 The paper “𝗔𝗻 𝗶𝗻𝘁𝗲𝗿𝗽𝗿𝗲𝘁𝗮𝗯𝗹𝗲 𝗺𝗮𝗰𝗵𝗶𝗻𝗲 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝘀𝘆𝘀𝘁𝗲𝗺 𝗳𝗼𝗿 𝗰𝗼𝗹𝗼𝗿𝗲𝗰𝘁𝗮𝗹 𝗰𝗮𝗻𝗰𝗲𝗿 𝗱𝗶𝗮𝗴𝗻𝗼𝘀𝗶𝘀 𝗳𝗿𝗼𝗺 𝗽𝗮𝘁𝗵𝗼𝗹𝗼𝗴𝘆 𝘀𝗹𝗶𝗱𝗲𝘀” was published in the international scientific journal npj Precision Oncology. 🔗 Read more about it here: https://lnkd.in/e4btHUZh This work is part of CADPATH.ai, an IMP Diagnostics project partially funded by the COMPETE 2020 programme - which focuses on the development of AI tools for colorectal and cervical pathologies.

Revolutionizing Prostate Cancer Diagnosis with AI! Attend an engaging talk by Dr. Alessandro Caputo, a pathologist passionate about digital pathology. In this Academy session, he explores the groundbreaking role of AI in improving both the diagnosis and treatment of prostate cancer. Learn how AI is transforming pathology workflows, making cancer detection faster and more precise! In this insightful session, you’ll learn: - The three main types of tissue samples pathologists work with: biopsies, transurethral resections (TURP), and prostatectomies—and how AI tools can optimize their analysis.  - How AI algorithms can improve accuracy in identifying prostate cancer while reducing the workload for pathologists.  - The challenges of distinguishing cancerous tissues from non-cancerous ones, and how AI can help avoid misdiagnosis. Whether you're a medical professional, a researcher, or simply curious about AI in healthcare, this lecture offers a deep dive into the future of prostate cancer diagnostics. Ready to explore how AI can reshape pathology? Join us at the ESDIP Academy and revolutionize your understanding of prostate cancer diagnosis with cutting-edge AI insights. Become an ESDIP member and access here: https://lnkd.in/dfTmZqB8 #ProstateCancer #AIInMedicine #PathologyInnovation #HealthcareTechnology #MedicalAI #FutureOfMedicine

Excited to share two of our groundbreaking studies at #ESTRO2024 that showcase the potential of AI and multi-omics data in transforming rectal cancer treatment, all done by the very talented Zhuoyan Shen! Part 1 - Sunday 4:30 PM, Hall 1 Introducing RADIANT: A proof-of-concept database linking multi-omics data from the ARISTOTLE trial for AI-driven predictive modelling in locally advanced rectal cancer (LARC). Key highlights: - Established a comprehensive database integrating clinical, radiotherapy, pathology, and genomic data from 589 LARC patients - Developed AI models to enrich the database, including automatic contouring of anatomical structures and analysis of tumor microenvironment - Enables hypothesis-driven and hypothesis-free analyses to personalize LARC treatment Part 2 - Monday 11:00 AM, Hall 2 A novel AI framework to quantify tumour infiltrating lymphocytes (TILs) and unravel their prognostic significance. Key findings: - Developed and validated an AI framework for automated TIL density quantification from whole slide images - High pre-treatment TIL density significantly associated with improved disease-free and overall survival outcomes - Radiotherapy-induced immune response observed in a subgroup of patients, correlating with better prognosis Join us at both sessions to learn more about these exciting advancements in #MedPhys, #Pathology, #RadOnc, and #AI! We look forward to engaging with you.

New paper from the Kather Lab: How deep learning identifies key mutations in colorectal cancer for better treatment options 👏 Congratulations to Marco Gustav, Zunamys Carrero, PhD, Chiara M. L. Loeffler and the other authors from the Clinical AI group of Jakob Nikolas Kather on their recent publication “Deep learning for dual detection of microsatellite instability and POLE mutations in colorectal cancer histopathology” in the Nature Portfolio journal ‘npj Precision Oncology’. 🔬 Key Highlights:                                                                              The study explores how AI and deep learning can predict specific genetic subtypes of colorectal cancer directly from histopathology slides. The workflow identifies colorectal cancer patients who might respond to immunotherapy, including those with microsatellite instability and rare POLE mutations, by only predicting microsatellite status from routine histology slides. This approach provides a fast, cost-effective pre-screening method, prioritizing cases for extensive sequencing.                                  🌐 The study is a collaborative effort of researchers from Université Paris-Est Créteil (UPEC), IMRB - Institut Mondor de Recherche Biomédicale, Greater Paris University Hospitals - AP-HP, Universität Augsburg, INSERM, Universitätsklinikum Carl Gustav Carus Dresden, Technische Universität Dresden, DKFZ Deutsches Krebsforschungszentrum, National Center for Tumor Diseases (NCT) Heidelberg, Universitätsklinikum Heidelberg (UKHD), University of Leeds Read the full publication here: https://lnkd.in/e_9-ShCa #DeepLearning #CancerResearch #PrecisionOncology #ColorectalCancer #AIinMedicine #MedicalInnovation #Immunotherapy

#foundationmodel, #cancer A pathology foundation model for cancer diagnosis and prognosis prediction tks for sharing #Isaac Kohane #Xiyue Wang,  #@ Junhan Zhao,  #Eliana Marostica,  #Yuan WEI,  #Aaron Jietian Jin,  #Jiayu Zhang,  #Ruijiang Li,  #@ Hongping Tang,  #@ Kanran Wang,  #@ Yu Li,  #@ Fang Wang,  #@ Yulong Peng,  #@ Junyou Zhu,  #@ Jing Zhang,  #@ Christopher Jackson,  #@ Jun Zhang,  #@ Deborah Dillon,  #@ Nancy U. Lin,  #Lynette Sholl,  #@ Thomas Denize,  #David Meredith,  #Keith Ligon,  #Sabina Signoretti,  #Shuji Ogino, … #Kun-Hsing Yu

Researchers developed CHIEF, an AI model that can diagnose and predict prognosis across 19 cancer types using pathology images. The model outperformed existing AI methods in tasks like cancer detection, origin identification, and outcome prediction, achieving up to 94% accuracy across multiple datasets. https://lnkd.in/gCBAtxsU #artificialintelligence #ai #cancer

From early detection to personalised care plans, AI is transforming the way we approach cancer diagnosis and treatment. “We are entering a new realm of data science” Professor Dow-Mu Koh, Consultant Radiologist in Functional Imaging, explains how he's using AI to drive advances in the hospital, and beyond: “A major project that I’ve been leading is using AI to assess the extent of bone disease in patients using whole-body MRI scans. The technology to process these images was developed with our commercial partner, Mint Medical, and is currently the first of its kind being tested in a multicentre trial. “This research could eventually help us identify whether treatment is working at a much earlier stage, allowing us to change treatments rapidly if needed and improve patient outcomes. While this study is focused on prostate cancer patients with bone disease, the software could be adapted for other tumour types, too. “Looking ahead, we are entering a new realm of data science. It could make the delivery of healthcare much more personalised and efficient, improving the patient experience as a result. “For example, we’re using AI to help acquire MRI data faster. In the near future, we could reduce the time a patient needs to spend on a scanner from one hour to just 15 minutes.” Discover more about our Integrated Pathology Unit: https://bit.ly/43Sr2xw

Researchers at Massachusetts Institute of Technology have developed an AI-driven real-time evaluation model for gastric cancer using a confocal laser endomicroscopic system. This innovative approach offers instantaneous histologic assessment and can serve as a complementary tool for pathologists. The AI model, which converts visual data into text for real-time detection, significantly enhances the accuracy of identifying cancer cells compared to traditional methods. By integrating AI assistance, pathologists' diagnostic performance markedly improves, showcasing the system's potential in clinical settings. This research could revolutionize intraoperative margin assessments and cancer diagnostics.  https://lnkd.in/gM-_u9aD #AI #MedTech #Medicine #VC #IP #Patents #DeepTech

Even a very basic model differentiating benign from malignant will save lots of time. I will go one step further, Even if we had a model to heatmap epithelial areas will be too much helpful. I recommend all friends to follow these series :)

We’re excited to announce the publication of the results from the external validation of Deep Bio Inc.’s DeepDx Prostate AI algorithm, conducted by Stanford University. Title: External Validation of An Artificial Intelligence Model for Gleason Grading of Prostate Cancer on Prostatectomy Specimens Published in: BJU Int 2024 Access the paper here: 10.1111/bju.16464 The validation study aimed to evaluate the algorithm’s performance for Gleason grading on whole-mount radical prostatectomy specimens. Initially trained on core needle biopsy (CNB) samples, the algorithm was tested on whole-mount RP specimens from an institution whose samples had not been previously encountered, without any prior tuning. Key Findings: - Cancer Detection: High agreement in identifying cancerous vs. non-cancerous tissue, with a quadratically weighted Cohen’s kappa (QWK) of 0.91. - Gleason Grading: Strong concordance in distinguishing individual Gleason grades, with QWK values of 0.89. Risk Group Classification: High precision in risk group classification, with a QWK of 0.89. Authors' Conclusions: The DeepDx Prostate AI algorithm is an accurate tool for identifying and grading prostate cancer on digital histopathology images of whole-mount RP specimens, demonstrating almost perfect concordance with expert GU pathologists and impressive performance in various clinically relevant tasks. These results underscore the AI's accuracy in cancer detection and Gleason grading, reinforcing its potential as a powerful tool in prostate cancer diagnostics. Note from Deep Bio Inc.: When used for its intended tissue type—CNB samples—our DeepDx Prostate AI algorithm demonstrates 99% sensitivity and 97% specificity. (Reference: Mod Pathol. 2022 Apr 29. doi: 10.1038/s41379-022-01077-9) Discover more about our specialized solution for RP tissue analysis on our website: https://lnkd.in/gRHfT2fd #DeepBio #DigitalPathology #AIinHealthcare