Us2.ai’s Post

Interesting study which shows that infrared, when augmented by AI, can be effective can be effective in screening for coronary artery disease. While an initial report, the results are impressive. What’s cool here is that the AI is enabling low cost and universally available technologies to take the place of high cost and high risk assessment tools. We’ll still need to verify disease severity with CT and treat with cath procedures, but now we’ve got something that can be used much more broadly. Put succinctly, this could be a real game changer: improve availability, enable early detection, and drive down costs.

This is an impressive application of AI that can elevate cardiovascular disease screening to a whole new level !👇 Traditionally, coronary artery diseases (CAD) are initially assessed by evaluating risk factors and symptoms (pre-test probability: PTP), which often have limited performance. This team used infrared thermography (IRT)—a technology that assesses surface temperature without skin contact—to predict CAD. The results are impressive: “The IRT image model demonstrated outstanding performance (AUC 0.804) compared to traditional PTP models (AUC 0.713).” This type of contactless technology, which is more accessible and far less invasive than CT and coronary angiography, combined with AI analysis, has the potential to significantly enhance the detection of cardiovascular diseases at scale. More broadly, AI's ability to extract clinically relevant information from common images that are not visible to human experts, as demonstrated here, opens up vast opportunities to improve the detection and analysis of human physiology and diseases. This is particularly true for radiology, which generates an immense amount of images every day, likely containing crucial information waiting to be unveiled by AI. Keep in mind that this study has several limitations, particularly the bias stemming from the inclusion of only high-risk CAD patients. Congrats to the team on this great study, and wishing everyone a fantastic start to the week ! #AI #thermography #facial #cardiology #CAD #screening Tsinghua University Link to the article in the comments 👇

🚨 A team in the Netherlands investigated the AliveCor #kardiamobile #AI algorithms in ⚡ 85,000 ECGs ⚡ of 2,300 patients in the #Hartwacht program of our partner Cardiologie Centra Nederland and came up with really interesting results! 👉 Determination "Sinus Rhyhtm": PPV of 99.6% 👉 Determination "Possible AFib": Sensitivity of 98.9% or in other words: 💥 if the algorithm says "Sinus Rhythm" it is highly likely true 💥 The algorithm is very sensitive to detect Afib The authors conclude: "Standalone algorithm interpretation of Sinus Rhythm ECGs by the KardiaMobile algorithm, without additional manual checking, is feasible." #mhealth #digitalhealth #cardiology Bridget Slaats Sebastiaan Blok G. Aernout Somsen, MD PhD FACC FESC Igor Tulevski Renoud Knops Bert-Jan van den Born Michiel Winter David Albert Ben Green, MD Link to the full publication: https://lnkd.in/d_j_F2bi

I’m thrilled to share that our work, “Classification, Regression, and Segmentation Directly from k-Space in Cardiac MRI,” has been accepted to the Machine Learning in Medical Imaging (MLMI 2024) workshop at #MICCAI. Cardiac Magnetic Resonance Imaging (CMR) is the gold standard for diagnosing cardiovascular diseases. Clinical diagnoses predominantly rely on magnitude-only Digital Imaging and Communications in Medicine (DICOM) images, omitting crucial phase information that might provide additional diagnostic benefits. In contrast, k-space is complex-valued and encompasses both magnitude and phase information, while humans cannot directly perceive. In this work, we propose KMAE, a Transformer-based model specifically designed to process k-space data directly, eliminating conventional intermediary conversion steps to the image domain. KMAE can handle critical cardiac disease classification, relevant phenotype regression, and cardiac morphology segmentation tasks. We utilize this model to investigate the potential of k-space-based diagnosis in cardiac MRI. Notably, this model achieves competitive classification and regression performance compared to image-domain methods e.g. Masked Autoencoders (MAEs) and delivers satisfactory segmentation performance with a myocardium dice score of 0.884. Last but not least, our model exhibits robust performance with consistent results even when the k-space is 8* undersampled. We encourage the MR community to explore the untapped potential of k-space and pursue end-to-end, automated diagnosis with reduced human intervention. Check out our work at https://lnkd.in/dmgatET6 This work stems from my Interdisciplinary Project in an Application Subject (IDP) at the AI in Medicine lab at #TUM, led by Prof. Daniel Rueckert. I am deeply grateful to my advisor, Jiazhen Pan, for his unwavering support and encouragement. Without him, I would not have had the courage to attempt the MICCAI conference in my third semester. Although we narrowly missed being accepted into the main conference, this experience has given me the courage and confidence to continue pursuing the AI+Medicine. Next time, we aim for the main conference! See you this October in #Morocco!

new ECG realizes the quantification of ECG for the first time. (a) As the waveform images are {Qualitative diagnostics}, the identification of patterns may lead to controversy between individuals. (b) The data are {Quantitative diagnostics}, and the data have international standards and will not cause disputes. Contains three main aspects of data, (1) Non-invasive ECG data. (2) Invasive electrophysiologic data. (3) The data in the textbook is calculated based on the SAN, because the frequency of the SAN is the heart rate 60~100/bpm. The ECG data is only accurate in the R~R interval because the R wave is at an acute angle. In the new ECG, each wave has an acute angle, and all measurements are in milliseconds (ms). Why should the ECG be digitized? (a) Many disease images do not change, but the data does. (b) Many images have changed and must be identified on the basis of the data. (c) For many non-occurring diseases, the image does not change, but the data changes first {For example: CAD, heart failure, AVNRT/AVRT, etc.} If you find them useful, worth reading, please comment, like and share. PhysioSign USA #ecg #noninvasive #ep #cardiology #electrophysiology #ai #electrocardiogram

🧠 **September 21st is World Alzheimer’s Disease Day** 🧠 In recognition of this important day, I would like to share an article I’ve been working on again: "Proposal for the Combined Use of Florbetapir in PET and Diffusion Tensor Imaging in Alzheimer’s Disease Diagnosis." Alzheimer’s Disease is a growing global challenge, and this paper explores a novel approach to improving early diagnosis by combining advanced imaging techniques, Positron Emission Tomography (PET) with Florbetapir, and Diffusion Tensor Imaging (DTI). By detecting both β-amyloid plaques and white matter integrity loss, this combined method holds promise for enhancing the accuracy of early diagnosis, which is critical for improving patient outcomes. While we still have much to learn, I hope this theoretical proposal sparks interest and collaboration in the field. It’s a small contribution to the ongoing efforts in understanding and combating this devastating disease. #AlzheimersDisease #Neuroimaging #MedicalResearch #WorldAlzheimersDay #HealthcareInnovation #PETImaging #DTI #Radiography #AlzheimersResearch #CognitiveNeuroscience #EarlyDiagnosis #BrainHealth #MRI

Great study using AI models to predict patient diagnosis and assess cardiac function, offering a non-invasive and efficient approach to cardiac care -> The study provides a non-invasive method to predict diagnosis of patients undergoing cardiac MRI and to obtain left ventricular end-diastolic pressure. -> The study identified 66936 patients with over 183772 individual left ventricular pressure measurements. -> The AI models classified the different types of cardiomyopathy with good accuracy. -> The AI models could be integrated into clinical practice and provide added value to the information content of cardiac MRI, allowing for disease classification and prediction of diastolic function. Read more on: https://lnkd.in/esFet85h The Lancet #digitalhealth #AI #cardiology

Note: Refer to Dr B. Hoffman and Dr A. Damato for mapping identification methods. Data values refer to the standard value of invasive electrophysiology. In animal experiments, body surface detection was performed under general anesthesia, the conventional ECG and the new ECG were detected simultaneously. As long as the ECG can be recorded, the new ECG can be recognized, and the scan recording rate is 99.99%. ***(The value of ST segment data is twice that of humans, and the reason is preferential research and exploration). The new ECG recorded new wavelets (sine waves that are smaller than the P and T waves, both of which occur in the time domain of the P and T waves) at both the atrium and ventricle. This swine 🐖 is very heavy, over 70 kilograms, so the atrial notch in lead II is very clear and the chest lead is a wave. The AVN has two tips, one of which is connected to the head of the bundle of His. *1~2 is atrial。2~3, 3~4, 4 ~5 are AVN waves。5~6 is His bundle。6~7 are bundle branches. Completely consistent with Hoffman's conjecture and the data values are within the standard range. If you enjoyed this article, please like and share! PhysioSign USA #ecg #noninvasive #ep #cardiology #electrophysiology #ai #electrocardiogram #innovation #medtech #heartdisease #heartattack #coronaryarterydisease #cardiovasculardisease

🥇 Congratulations to Dr @Ramesh Nadarajah, winner of the Heart Rhythm Congress (HRC) 2024 Young Investigators Competition for Clinical Science! This prestigious competition recognises emerging researchers presenting cutting-edge work in arrhythmia care. Dr Nadarajah’s winning presentation, ‘A Federated Data Analytic and Meta-Machine Learning Approach to Predict Incident Atrial Fibrillation in Patients with Acute Stroke: FIND-AF STROKE’, showcases a landmark analysis in stroke care. The study demonstrates how federated data analytics and meta-machine learning can create a generalisable decision support tool to accurately identify patients needing extended AF monitoring after a stroke, with real-world impact across diverse populations. Read the abstract here: https://loom.ly/3-VGLp8 #MedicalResearch #HealthScience #Cardiology #Arrhythmia #Stroke

🎉🫀 We are thrilled to announce that our abstract, "A smartphone-based AI model to detect left ventricular systolic dysfunction on 12-lead ECG," has been accepted for presentation at the ESC Congress 2024 in London and will be presented at the Digital Health Stage! 💙 Stay tuned for more updates on our activities with #PMcardio at #ESC2024! #PowerfulMedical #CardiologyAI #MedicalResearch