"HybridNet: A Novel Approach for Myocardial Infarction Segmentation"

📢 From @BME_Horizon A Multi-Modal Medical Image Fusion Framework Based on Two-Scale Decomposition and Saliency Detection 🧠🖼️ Authors: Harmanpreet Kaur, Renu Vig, Naresh Kumar Abstract: This research proposes a novel fusion approach for multimodal medical images utilizing mean filtering and maximum symmetric surround saliency detection. The method decomposes images into base and detail layers, fuses the detail layers using guided filtering, and employs linear summation for the base layer. The fused images demonstrate superior performance in visual appearance and quantitative assessment compared to existing methods, effectively preserving edges and energy information across modalities. Key Findings: ✅ 𝗜𝗺𝗮𝗴𝗲𝘀 𝗱𝗲𝗰𝗼𝗺𝗽𝗼𝘀𝗲𝗱 into 𝗯𝗮𝘀𝗲 𝗮𝗻𝗱 𝗱𝗲𝘁𝗮𝗶𝗹 𝗹𝗮𝘆𝗲𝗿𝘀 𝘂𝘀𝗶𝗻𝗴 𝗺𝗲𝗮𝗻 𝗳𝗶𝗹𝘁𝗲𝗿𝗶𝗻𝗴. ✅ 𝗗𝗲𝘁𝗮𝗶𝗹 𝗹𝗮𝘆𝗲𝗿𝘀 fused via guided filtering; base layers fused using 𝗹𝗶𝗻𝗲𝗮𝗿 𝘀𝘂𝗺𝗺𝗮𝘁𝗶𝗼𝗻. ✅ Evaluated on a 𝘄𝗶𝗱𝗲𝗹𝘆 𝘂𝘀𝗲𝗱 𝗺𝗲𝗱𝗶𝗰𝗮𝗹 𝗶𝗺𝗮𝗴𝗲 𝗱𝗮𝘁𝗮𝗯𝗮𝘀𝗲 with 𝗶𝗺𝗽𝗿𝗼𝘃𝗲𝗱 𝗽𝗲𝗿𝗳𝗼𝗿𝗺𝗮𝗻𝗰𝗲. ✅ 𝗘𝗳𝗳𝗲𝗰𝘁𝗶𝘃𝗲𝗹𝘆 𝗽𝗿𝗲𝘀𝗲𝗿𝘃𝗲𝘀 𝗲𝗱𝗴𝗲𝘀 and energy information in fused images. ✅ Enhances accuracy and reliability for diagnostic imaging and 𝗶𝗻𝘁𝗿𝗮-𝗼𝗽𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗴𝘂𝗶𝗱𝗮𝗻𝗰𝗲. Full study available here:  https://lnkd.in/gBsHXAmJ #MedicalImaging #ImageFusion #MultimodalImaging #SaliencyDetection #PixelLevelFusion #HealthcareAI #ComputerVision

🔬 𝐓𝐡𝐫𝐢𝐥𝐥𝐞𝐝 𝐭𝐨 𝐒𝐡𝐚𝐫𝐞 𝐎𝐮𝐫 𝐋𝐚𝐭𝐞𝐬𝐭 𝐈𝐄𝐄𝐄 𝐑𝐞𝐬𝐞𝐚𝐫𝐜𝐡 𝐏𝐮𝐛𝐥𝐢𝐜𝐚𝐭𝐢𝐨𝐧! I’m excited to announce that our research paper, “𝐄𝐟𝐟𝐢𝐜𝐢𝐞𝐧𝐭 𝐁𝐫𝐚𝐢𝐧 𝐓𝐮𝐦𝐨𝐫 𝐃𝐞𝐭𝐞𝐜𝐭𝐢𝐨𝐧 𝐰𝐢𝐭𝐡 𝐚 𝐋𝐢𝐠𝐡𝐭𝐰𝐞𝐢𝐠𝐡𝐭 𝐂𝐮𝐬𝐭𝐨𝐦𝐢𝐳𝐞𝐝 𝐂𝐍𝐍 𝐀𝐩𝐩𝐫𝐨𝐚𝐜𝐡,” has been published in the IEEE Xplore Digital Library as part of the 2025 𝐈𝐧𝐭𝐞𝐫𝐧𝐚𝐭𝐢𝐨𝐧𝐚𝐥 𝐂𝐨𝐧𝐟𝐞𝐫𝐞𝐧𝐜𝐞 𝐨𝐧 𝐌𝐢𝐜𝐫𝐨𝐰𝐚𝐯𝐞, 𝐎𝐩𝐭𝐢𝐜𝐚𝐥, 𝐚𝐧𝐝 𝐂𝐨𝐦𝐦𝐮𝐧𝐢𝐜𝐚𝐭𝐢𝐨𝐧 𝐄𝐧𝐠𝐢𝐧𝐞𝐞𝐫𝐢𝐧𝐠 (𝐈𝐂𝐌𝐎𝐂𝐄 𝟐𝟎𝟐𝟓). 📌 𝐊𝐞𝐲 𝐇𝐢𝐠𝐡𝐥𝐢𝐠𝐡𝐭𝐬: 🧠 We developed a lightweight customized CNN model that achieves an impressive 98.03% classification accuracy with significantly reduced computational overhead. ⚙️ Ideal for real-time applications and resource-constrained environments, including mobile healthcare and edge devices. 🩺 Our work pushes the boundary of AI-driven medical diagnostics, especially in early-stage brain tumor detection. A big thank you to my co-authors — Shantanu Shookdeb, Purnima Pattnaik, Dr. DEBENDRA MUDULI, and Dr. SANTOSH KUMAR SHARMA — for their incredible teamwork and contributions throughout this journey! 🔗 𝐑𝐞𝐚𝐝 𝐭𝐡𝐞 𝐟𝐮𝐥𝐥 𝐩𝐚𝐩𝐞𝐫 𝐡𝐞𝐫𝐞: https://lnkd.in/gjFqTshA #DeepLearning #CNN #BrainTumorDetection #MedicalImaging #AIinHealthcare #IEEE #ICMOCE2025 #ResearchPublication #Teamwork #ProudMoment

I am pleased to share that my research paper, “𝗧𝗿𝗶𝗩𝗲𝗿𝘀𝗲𝗡𝗲𝘁: 𝗠𝘂𝗹𝘁𝗶-𝗕𝗿𝗮𝗻𝗰𝗵 𝗥𝗲𝘀𝗡𝗲𝘁 𝗙𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸 𝗳𝗼𝗿 𝗘𝗻𝗵𝗮𝗻𝗰𝗲𝗱 𝗘𝗻𝗱𝗼𝘀𝗰𝗼𝗽𝗶𝗰 𝗟𝗲𝘀𝗶𝗼𝗻 𝗖𝗹𝗮𝘀𝘀𝗶𝗳𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝘃𝗶𝗮 𝗘𝗱𝗴𝗲 𝗮𝗻𝗱 𝗧𝗲𝘅𝘁𝘂𝗿𝗲 𝗜𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻”, has now been published in 𝗜𝗘𝗘𝗘 𝗫𝗽𝗹𝗼𝗿𝗲. This work introduces 𝗧𝗿𝗶𝗩𝗲𝗿𝘀𝗲𝗡𝗲𝘁, a multi-branch deep learning framework that combines 𝗰𝗼𝗹𝗼𝗿, 𝗲𝗱𝗴𝗲, 𝗮𝗻𝗱 𝘁𝗲𝘅𝘁𝘂𝗿𝗲 𝗿𝗲𝗽𝗿𝗲𝘀𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻𝘀 to improve lesion classification in endoscopic images. By designing a ResNet-based architecture with feature-level integration, the model achieves more robust performance compared to conventional single-branch approaches. The motivation behind this research stems from the critical role of 𝗰𝗼𝗺𝗽𝘂𝘁𝗲𝗿-𝗮𝗶𝗱𝗲𝗱 𝗱𝗶𝗮𝗴𝗻𝗼𝘀𝘁𝗶𝗰𝘀 𝗶𝗻 𝗴𝗮𝘀𝘁𝗿𝗼𝗶𝗻𝘁𝗲𝘀𝘁𝗶𝗻𝗮𝗹 𝗱𝗶𝘀𝗲𝗮𝘀𝗲 𝗱𝗲𝘁𝗲𝗰𝘁𝗶𝗼𝗻, where accurate classification can support early diagnosis and clinical decision-making. Our study contributes to advancing 𝗔𝗜 𝗶𝗻 𝗵𝗲𝗮𝗹𝘁𝗵𝗰𝗮𝗿𝗲, with an emphasis on practical improvements in 𝗺𝗲𝗱𝗶𝗰𝗮𝗹 𝗶𝗺𝗮𝗴𝗲 𝗮𝗻𝗮𝗹𝘆𝘀𝗶𝘀. I am deeply grateful to my guiding faculty Kimmi Gupta for her constant mentorship, and to my teammates Dhruv Chahar and Anmol Pangtey for their collaboration and support throughout this work. The full paper is available on IEEE Xplore: https://lnkd.in/g9YvJkfr I would also like to thank Bennett University and the School of CSET Bennett University, India for providing the platform and opportunity to pursue this research. #DeepLearning #ComputerVision #MedicalImaging #Endoscopy #LesionClassification #AIinHealthcare #IEEE #ResearchPublication #MedicalAI #NETCRYPT2025

Our Patent Application titled “AI-Powered Stacking Ensemble Model for EEG Signal Classification in BCI System” has been officially published in The Patent Office Journal No. 35/2025, dated 29/08/2025 (Application No. 202541077335 A). This innovation, contributed by our team — 👨🔬 Srinivas Rao Gorre 👨🔬 Dr. Ravichander Janapati 👨🔬 Ch. Rajendra Prasad introduces a next-generation EEG-based Brain-Computer Interface (BCI) system powered by advanced AI techniques, featuring: 🔹 Stacking ensemble classification combining multiple ML models 🔹 Extraction of time, frequency, and power spectral features 🔹 Data augmentation (Gaussian noise, time-shifting, amplitude scaling) for robust training 🔹 A validation module ensuring adaptability across individuals 🔹 Real-time signal processing for BCI applications This breakthrough aims to significantly enhance EEG signal classification accuracy, enabling applications in neuro-rehabilitation, assistive technologies, and human-computer interaction. Proud to see our research translating into impactful innovations! 🚀 #Innovation #Patent #ArtificialIntelligence #MachineLearning #EnsembleLearning #EEG #BCI #Neurotechnology #SignalProcessing #Research #HealthcareAI #SRUniversity #ProudMomentSRSRSR UniversityECE SR UniversityCS&AI -SRU

🩻 The Power of Bioimaging in Modern medicine Bioimaging has become one of the most transformative tools in healthcare. From MRI and CT scans to cutting-edge molecular and optical imaging, these technologies allow us to see inside the body with unprecedented clarity. But the real shift is happening at the intersection of imaging and artificial intelligence. Machine learning algorithms are now able to detect patterns invisible to the human eye, assisting clinicians in identifying diseases earlier, guiding treatment decisions, and monitoring outcomes with greater precision. Bioimaging is more than just pictures — it is the bridge between engineering, data, and medicine, enabling a deeper understanding of health and disease. As innovation continues, the future holds even more powerful imaging solutions that will make diagnostics faster, smarter, and more accessible. #Bioimaging #BiomedicalEngineering #AIinHealthcare #MedicalInnovation #HealthcareTechnology

Just wrapped up a fascinating computer vision project on classifying medical images. Our team built a hybrid model for skin lesion classification that combines classical features with modern Vision Transformer (ViT) embeddings. The biggest challenge? The highly imbalanced dataset, a common issue in medical imaging. We tackled this with oversampling and class weighting, but the real breakthrough came from probability-threshold tuning. By masking low-confidence predictions, we saw a significant jump in accuracy to 93.8% and a Macro-F1 score of 0.8572, making the model's outputs much more reliable. This project highlights how combining classical computer vision features with modern transformer-based embeddings can improve performance and generalization in high-stakes domains like healthcare. This approach also extends to other imbalanced domains such as insurance and finance. Read the full presentation to learn more about our methodology and findings! A huge thanks to my incredible teammates #ConorHuh, #LynnWang, #MridulJain, and #SenthilPeriasamy and #VashaDutell for their expert based insights. #ComputerVision #MachineLearning #UCBerkeley #ViT #DataScience #AIHealthcare

🗺️ Spatial Multiplexing of Whispering Gallery Mode Sensors 🧑 Stephen Holler* and Matthew Speck 🏫 Fordham University 🔎 #Whisperinggallerymode #resonators have proven to be robust and sensitive platforms for the trace detection of chemical and/or biological analytes. Conventional approaches using serially addressed resonators face challenges in simultaneous multi-channel (i.e., multi-species) detection. We present an alternative monitoring scheme that allows for the spatial multiplexing of whispering gallery mode resonators with the simultaneous observation of the resonance spectra from each of them. By imaging arrays of microspheres and monitoring the glare spot intensities through image processing routines, resonance spectra from multiple resonators may be simultaneously recorded without interference or confounding effects of serial excitation/detection. We demonstrate our multiplexed imaging approach with bulk refractive index variations and virus–antibody binding. https://lnkd.in/gTeGv-Z9

🔬 Exciting Publication Alert! 🔬 We are delighted to share that our FS-13 hyperspectral imaging system (FS-1X Series) was used in a recent study published in the IEEE Sensors Journal (2025) Researchers from Chongqing University of Posts and Telecommunications and Army Medical University developed CBIA-Net, a novel deep learning framework, to achieve rapid and non-destructive detection of wound bacteria. Using hyperspectral images captured by the FS-13, they built a large dataset covering seven typical wound pathogens at multiple concentrations and achieved 97% accuracy, precision, and recall in bacterial classification. This research demonstrates the powerful potential of CHNSpec’s FS-1X Series in biomedical imaging and clinical diagnostics, extending the applications of hyperspectral technology far beyond traditional fields. 👏 Congratulations to the research team for this innovative achievement, and thank you for trusting CHNSpec technology! #HyperspectralImaging #BiomedicalResearch #ClinicalDiagnostics #CHNSpec

It always amazes me truly how long it takes to communicate about innovation until solid clincal data and patient outcomes actually change daily practice and consultation pathways. Rayner’s 2025 ESCRS Symposium was no doubt the most seeked scientific event this year - no other topic but AI spiral optics and outcomes set the trend. This technology has it all: a degree of sexiness, data robustness, a degree of cheekiness to challenge the status quo and clinical outcomes to win every surgeons in a storm ❤️

📢 The Special Issue "New Sights of Machine Learning and Digital Models in Biomedicine" is open for submissions! 🥼 This Special Issue is guest-edited by Prof. Dr. Hatem Alhadainy. 💡 This Special Issue explores the intersection of machine learning (ML), digital modeling, and biomedicine, highlighting innovative approaches that leverage advanced computational techniques to enhance medical research, diagnosis, treatment, and patient care. 🔗 Click the link to access more details about the Special Issue: https://lnkd.in/gxs_Gybe 🕑 The deadline for manuscript submissions is 31 January 2026. 🎉 Welcome to join us as authors and reviewers! 👏 And welcome to follow our LinkedIn account @Bioengineering MDPI. #Machine_Learning #Digital_Models #Biomedicine