Neuromechanics Research Lab (NRL) IIT Delhi’s Post

Supervised vs. Unsupervised  Learning This framework is a helpful guide for selecting the appropriate algorithm based on the data type and research objective (example for biomedical studies) 🙂

🎉 Excited to announce the publication of our paper: “Deep Learning-Based Classification of Echocardiographic Perspectives: Emphasis on Parasternal Long Axis and Apical 4 Chamber for Mitral Valve Assessment,” presented at the 2023 2nd International Engineering Conference on Electrical, Energy, and Artificial Intelligence (EICEEAI) at IEEE Xplore. In this research, we leveraged advanced deep learning techniques to enhance the diagnostic capabilities of echocardiography, particularly focusing on the Parasternal Long Axis (PLA) and Apical 4 Chamber (A4C) views. By utilizing convolutional neural networks and sophisticated image analysis methods, we aimed to improve the accuracy and sensitivity of mitral valve disease detection. Our study presents a novel approach to classifying echocardiographic perspectives, which can potentially revolutionize how medical professionals assess and diagnose mitral valve conditions. The integration of deep learning in medical imaging holds significant promise for increasing diagnostic precision and reducing operator dependency. Special thanks to the incredible team Ahmed Hamdy, Reem Hesham, Philo Magdy, Hana Saleh, and our dedicated and beloved supervisor, Assoc. Prof. Mohanad deif , PhD for their invaluable support and contributions. This achievement would not have been possible without their hard work and dedication. Additionally, I extend my appreciation to Prof. Dr. Rania AbdElrahman Elgohary, Dean of the College, for her continuous support and encouragement. Read more about our research here: https://lnkd.in/d6jGfTUv #DeepLearning #Echocardiography #MedicalImaging #Research #ComputerVision

Our research, titled "Rapid estimation of left ventricular contractility with a physics-informed neural network inverse modeling approach," is now published in Artificial Intelligence in Medicine. In this work, we propose a rapid framework for estimating heart function. We trained a physics-informed neural network (PINN) to learn the forward simulation by minimizing the residuals of a system of differential equations governing blood circulation. The real-time forward simulation enabled by the trained neural network allows us to efficiently solve the inverse parameter estimation problem. We showed that the inverse modeling using the trained PINN takes less than 3 minutes per case, compared to up to 60 hours with traditional numerical methods. PINN offers a promising solution to the time-consuming limitations of traditional numerical methods for parameter estimation and inverse problems. 🔗Full paper (Free until December 11, 2025): https://lnkd.in/etq8e8mT 🔗Code: https://lnkd.in/ezpJF3NQ

GraphBNC: Predicting metal nanocluster-protein interactions with machine learning Understanding how metal nanocluster interact with proteins is crucial for developing cutting-edge applications in bioimaging, biosensing, and nanomedicine. However, modeling these interactions poses challenges due to the complexity of the nano-bio interface. In a recent study, Pihlajamäki and colleagues introduced GraphBNC, a machine learning-based tool designed to predict interactions between water-soluble gold nanoclusters (AuNCs) and blood proteins. GraphBNC leverages graph theory to represent proteins, with protein residues modeled as nodes. Using neural networks, the tool predicts the strength of interactions with gold nanoclusters based on features such as chemical composition and spatial structure. The feedforward neural networks are trained on data from molecular dynamics (MD) simulations, which capture detailed atom-scale interactions between AuNCs and proteins. To further optimize predictions, GraphBNC incorporates a Monte Carlo-based simulated annealing process, exploring possible configurations and minimizing energy to identify the most likely interaction sites. The model’s predictions were validated through 500-nanosecond MD simulations, which demonstrated that the predicted binding sites remained stable for over 90% of the simulation time, underscoring the robustness of the model. By accurately predicting protein-gold nanocluster interactions, GraphBNC has the potential to accelerate the development of targeted drug delivery systems, biosensors, and other biomedical applications—particularly in cases where experimental data is scarce. Paper: https://lnkd.in/dXQDcSqZ #MachineLearning #Nanotechnology #Bioimaging #Biosensing #Nanomedicine #GoldNanoclusters #AI #ComputationalBiology #MolecularDynamics #GraphTheory #BiomedicalInnovation #DrugDelivery #Biosensors #Bioinformatics

☑️ *READ ASTRACT BELOW:* Keywords: Aneurysm; Deep learning; Magnetic resonance angiography; Medical image processing; Segmentation. Three-dimensional vessel model reconstruction from patient-specific magnetic resonance angiography (MRA) images often requires some manual maneuvers. This study aimed to establish the deep learning (DL)-based method for vessel model reconstruction. Time of flight MRA of 40 patients with internal carotid artery aneurysms was prepared, and three-dimensional vessel models were constructed using the threshold and region-growing method. Using those datasets, supervised deep learning using 2D U-net was performed to reconstruct 3D vessel models. The accuracy of the DL-based vessel segmentations was assessed using 20 MRA images outside the training dataset. The dice coefficient was used as the indicator of the model accuracy, and the blood flow simulation was performed using the DL-based vessel model. The created DL model could successfully reconstruct a three-dimensional model in all 60 cases. The dice coefficient in the test dataset was 0.859. Of note, the DL-generated model proved its efficacy even for large aneurysms (> 10 mm in their diameter). The reconstructed model was feasible in performing blood flow simulation to assist clinical decision-making. Our DL-based method could successfully reconstruct a three-dimensional vessel model with moderate accuracy. Future studies are warranted to exhibit that DL-based technology can promote medical image processing. Koizumi S, Med Biol Eng Comput. 2024 Oct;62(10):3225-3232. doi: 10.1007/s11517-024-03136-6. Epub 2024 May 28. PMID: 38802608; PMCID: PMC11379798. #Gesundheit #Bildung #Fuehrung #Coaching #Mindset #Motivation #Gehirn #Neuroscience #Psychologie #Persoenlichkeitsentwicklung #Kindheit #KeyNoteSpeaker #Humangenetik #Biochemie #Neuroleadership #Ernaehrung #Transformation #Stress #Demografie #Gender #Age #interkulturelleKompetenz #Epigenetik #Veraenderung #EmotionaleIntelligenz #Change #Gesellschaft #Organisationsentwicklung #Philosophie #Beratung # Quantum

The following is a good synopsis of the state of advanced imaging techniques and the effect of #AI on advanced image generators in life sciences. #cryoem #cryoet #sem #tem #lifesciences https://lnkd.in/ge9mjVKg

🔬✨ CT-Less Organ Segmentation in PET: A Deep Learning Preprint ✨🔬 We’re excited to share a preprint led by Yazdan Salimi from Geneva University Hospital, under supervision from Prof. Habib Zaidi, and Dr. Ismini Mainta, proposing a groundbreaking approach to multi-organ segmentation in PET imaging without the need for CT. 🧠📊 🔹 Study Overview: Traditional segmentation in PET imaging relies heavily on CT, but CT-PET alignment issues and PET-only workflows can limit its effectiveness. This study introduces a CT-independent deep learning model using PET emission data alone, addressing these challenges head-on. Models and Data: Advanced nnU-Net models were trained and tested on a dataset of over 2,000 PET/CT images, focusing on 18F-FDG and 68Ga-PSMA tracers. Innovation: By eliminating CT reliance, this pipeline enables consistent, accurate organ segmentation, even in PET-only environments. 🚀 🔹 Key Findings: High Accuracy: Achieved Dice coefficients of 0.81–0.82 for 18F-FDG and 0.77–0.79 for 68Ga-PSMA, with strong performance in organs like the brain and lungs. Versatility: Supports applications in dosimetry, kinetic modeling, and radiomics, expanding the utility of PET imaging in clinical and research settings. GitHub Access: The authors have made their repository publicly available for exploration and feedback — try it out here: 🔗 https://lnkd.in/ejhkigVx 👏 We wish this preprint success in finding a peer-reviewed journal for publication and invite the community to explore and engage with this exciting new approach! 🔗 Read the preprint: https://lnkd.in/eC9HkqNg #NuclearMedicine #DeepLearning #PETImaging #MedicalImaging #ArtificialIntelligence #OrganSegmentation #SwissResearch #OpenScience

🧠 Organoid Intelligence: How the FinalSpark Neuroplatform is Pushing the Boundaries of Biocomputing! 🔬The emerging field of organoid intelligence (OI) represents the next frontier in computing. At its core, OI leverages the power of brain organoids —lab-grown, 3D cultures of human brain cells—to process information and perform computational tasks, offering capabilities far beyond conventional AI. This innovation is spearheaded by companies like FinalSpark, whose “Neuroplatform” offers remote access to these living bioprocessors, allowing researchers to tap into the potential of biocomputing from anywhere in the world. 🖥️ The Potential of Biocomputing: Traditional computers rely on silicon processors, which, while powerful, require significant energy to perform tasks. In contrast, a human brain can execute complex computations with remarkable energy efficiency, using only around 20 watts of power to run 86 billion neurons. FinalSpark's bioprocessors emulate this efficiency by harnessing the computational power of brain organoids, potentially reducing energy consumption by a million times compared to traditional processors. 🧬 The applications for OI are vast, from revolutionizing artificial intelligence and machine learning to improving our understanding of neurological conditions like Alzheimer’s. OI could also give rise to new kinds of neuromorphic computing systems, mimicking the brain's ability to learn, adapt, and process information in real-time. ⚡️FinalSpark's Neuroplatform: FinalSpark's Neuroplatform offers researchers the ability to conduct 24/7 experiments on 16 human brain organoids via an integrated R&D environment. With real-time neural stimulation and reading capabilities, researchers can remotely test the functionality and intelligence of these organoids, furthering biocomputing innovation. ⚖️ Ethical Considerations: While the promise of OI and biocomputing is exciting, it also raises profound ethical questions. The ability to create living, thinking systems from human cells could lead to concerns about consciousness, autonomy, and the manipulation of living tissue for computational purposes. As these systems become more advanced, defining what constitutes "intelligence" and whether organoids possess any rights or consciousness will need careful consideration. #OrganoidIntelligence #Biocomputing #Neuroplatform #EthicalAI #WetwareComputing #AIInnovation #Neurotechnology #FinalSpark #EnergyEfficiency #FutureOfComputing #OI #AI

I am thrilled to share that another research paper of mine has been published in the #Medical Engineering & Physic journal. #Elsevier publication #ScopusQ1 journal, #SCI with an impressive #HIndex of 125. 📄 Title: "Automatic Diagnosis of Epileptic Seizures using Entropy-based Features and Multimodel Deep Learning Approaches" This research represents a significant step forward in the field of medical engineering, aiming to enhance the automatic diagnosis of epileptic seizures through innovative entropy-based features and advanced deep learning models. I extend my deepest gratitude to my co-authors, mentors, and everyone who supported this research journey. Your encouragement and guidance were invaluable. Looking forward to sharing our findings with the broader scientific community and contributing to advancements in medical technology. Feel free to check out the publication and share your thoughts! https://lnkd.in/gEPZzQ4i #Research #MedicalEngineering #DeepLearning #Epilepsy #AI #MachineLearning #ScientificResearch #AcademicPublishing #Gratitude #TeamWork

Last week, I successfully defended my master's thesis on 'Analysis of infrared thermal images for the study of diabetic foot disease', thus completing the Master in Advanced Telecommunications Technologies (MATT) from UPC - ETSETB Telecos BCN. This thesis has been developed under the umbrella of a collaborative project between IDIAPJGol (Institut d'Investigació en Atenció Primària de Salut Jordi Gol) and Bellvitge Biomedical Research Institute - IDIBELL to evaluate infrared thermal (IRT) images as an alternative method for the primary care settings to detect high-risk diabetic foot cases. Even though the preliminary results are not promising, I'm really proud of the work done to implement the complete image preprocessing and analysis system, which required some image processing and computer vision techniques. Overall, this master has permitted me to extend my knowledge about AI topics, from Machine Learning to LLMs, and has helped me to launch my professional career in the AI world. Let's keep going! #AI #ComputerVision #ImageProcessing #Biomedical