Interpretable and multimodal fusion methodology to predict severe hypoglycemia in adults with type 1 diabetes

Engineering Applications of Artificial Intelligence, Год журнала: 2025, Номер 144, С. 110142 - 110142

Опубликована: Янв. 30, 2025

Язык: Английский

---

Novelty Classification Model Use in Reinforcement Learning for Cervical Cancer

Cancers, Год журнала: 2024, Номер 16(22), С. 3782 - 3782

Опубликована: Ноя. 10, 2024

Cervical cancer significantly impacts global health, where early detection is piv- otal for improving patient outcomes. This study aims to enhance the accuracy of cervical diagnosis by addressing class imbalance through a novel hybrid deep learning model.

Язык: Английский

---

Force Map-Enhanced Segmentation of a Lightweight Model for the Early Detection of Cervical Cancer

Diagnostics, Год журнала: 2025, Номер 15(5), С. 513 - 513

Опубликована: Фев. 20, 2025

Background/Objectives: Accurate and efficient segmentation of cervical cells is crucial for the early detection cancer, enabling timely intervention treatment. Existing models face challenges with complex cellular arrangements, such as overlapping indistinct boundaries, are often computationally intensive, which limits their deployment in resource-constrained settings. Methods: In this study, we introduce a lightweight model specifically designed cell analysis. The employs MobileNetV2 architecture feature extraction, ensuring minimal parameter count conducive to real-time processing. To enhance boundary delineation, propose novel force map approach that drives pixel adjustments inward toward centers cells, thus improving separation densely packed areas. Additionally, integrate extreme point supervision refine outcomes using annotations, rather than full pixel-wise labels. Results: Our was rigorously trained evaluated on comprehensive dataset images. It achieved Dice Coefficient 0.87 Boundary F1 Score 0.84, performances comparable those advanced but considerably lower inference times. optimized operates at approximately 50 frames per second standard low-power hardware. Conclusions: By effectively balancing accuracy computational efficiency, our addresses critical barriers widespread adoption automated tools. Its ability perform real time low-cost devices makes it an ideal candidate clinical applications low-resource environments. This advancement holds significant potential enhancing access cancer screening diagnostics worldwide, thereby supporting broader healthcare initiatives.

Язык: Английский

---

Boosting adversarial transferability in vision-language models via multimodal feature heterogeneity

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Март 2, 2025

Vision-language pre-training models have achieved significant success in the field of medical imaging but exhibited vulnerability to adversarial examples. Although attacks are harmful, they valuable revealing weaknesses VLP and enhancing their robustness. However, due under-utilization modal differences consistent features existing methods, attack effectiveness migration samples not satisfactory. To address this issue enhance transferability, we propose multimodal feature heterogeneous framework. capability, a heterogenization method based on triplet contrastive learning, utilizing data augmentation cross-modal global intra-modal global-local mutual information learning. This further heterogenizes between modalities into distinct features, thereby improving capability. improve variance aggregation-based multi-domain perturbation method, using text-guided image perturb spatial frequency while combining previous gradient momentum, achieving better transferability. Extensive experiments demonstrate MFHA's advantage transferable with an average improvement 16.05%, outstanding performance large language like MiniGPT4 LLaVA. The work did has been open-sourced GitHub: https://github.com/doyoudooo/MFHA .

Язык: Английский

---

Challenging the status quo: Why artificial intelligence models must go beyond accuracy in cervical cancer diagnosis

Biomedical Signal Processing and Control, Год журнала: 2025, Номер 105, С. 107620 - 107620

Опубликована: Фев. 3, 2025

Язык: Английский

---

RL-Cervix.Net: A Hybrid Lightweight Model Integrating Reinforcement Learning for Cervical Cell Classification

Diagnostics, Год журнала: 2025, Номер 15(3), С. 364 - 364

Опубликована: Фев. 4, 2025

Background: Reinforcement learning (RL) represents a significant advancement in artificial intelligence (AI), particularly for complex sequential decision-making challenges. Its capability to iteratively refine decisions makes it ideal applications medicine, such as the detection of cervical cancer; major cause mortality among women globally. The Pap smear test, crucial diagnostic tool cancer, benefits from enhancements AI, facilitating development automated systems that improve screening effectiveness. This research introduces RL-Cervix.Net, hybrid model integrating RL with convolutional neural network (CNN) technologies, aimed at elevating precision and efficiency cancer screenings. Methods: RL-Cervix.Net combines robust ResNet-50 architecture reinforcement module tailored unique challenges cytological image analysis. was trained validated using three extensive public datasets ensure its effectiveness under realistic conditions. A novel application dynamic feature refinement adjustment based on reward functions employed optimize capabilities model. Results: innovative integration into CNN framework allowed achieve an unprecedented classification accuracy 99.98% identifying atypical cells indicative lesions. demonstrated superior interpretability compared existing methods, addressing variability complexities inherent images. Conclusions: marks breakthrough AI medical diagnostics, early cancer. By significantly improving efficiency, has potential enhance patient outcomes through earlier more precise identification disease, ultimately contributing reduced rates improved healthcare delivery.

Язык: Английский

---

The Diagnostic Classification of the Pathological Image Using Computer Vision

Algorithms, Год журнала: 2025, Номер 18(2), С. 96 - 96

Опубликована: Фев. 8, 2025

Computer vision and artificial intelligence have revolutionized the field of pathological image analysis, enabling faster more accurate diagnostic classification. Deep learning architectures like convolutional neural networks (CNNs), shown superior performance in tasks such as classification, segmentation, object detection pathology. has significantly improved accuracy disease diagnosis healthcare. By leveraging advanced algorithms machine techniques, computer systems can analyze medical images with high precision, often matching or even surpassing human expert performance. In pathology, deep models been trained on large datasets annotated pathology to perform cancer diagnosis, grading, prognostication. While approaches show great promise challenges remain, including issues related model interpretability, reliability, generalization across diverse patient populations imaging settings.

Язык: Английский

---

Pixel level deep reinforcement learning for accurate and robust medical image segmentation

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Март 10, 2025

Existing deep learning methods have achieved significant success in medical image segmentation. However, this largely relies on stacking advanced modules and architectures, which has created a path dependency. This dependency is unsustainable, as it leads to increasingly larger model parameters higher deployment costs. To break dependency, we introduce reinforcement enhance segmentation performance. current face challenges such high training cost, independent iterative processes, uncertainty of masks. Consequently, propose Pixel-level Deep Reinforcement Learning with pixel-by-pixel Mask Generation (PixelDRL-MG) for more accurate robust PixelDRL-MG adopts dynamic update policy, directly segmenting the regions interest without requiring user interaction or coarse We Asynchronous Advantage Actor-Critic (PA3C) strategy treat each pixel an agent whose state (foreground background) iteratively updated through direct actions. Our experiments two commonly used datasets demonstrate that achieves superior performances than state-of-the-art baselines (especially boundaries) using significantly fewer parameters. also conducted detailed ablation studies understanding facilitate practical application. Additionally, performs well low-resource settings (i.e., 50-shot 100-shot), making ideal choice real-world scenarios.

Язык: Английский

---

Polyp segmentation in medical imaging: challenges, approaches and future directions

Artificial Intelligence Review, Год журнала: 2025, Номер 58(6)

Опубликована: Март 17, 2025

Язык: Английский

---

Enhancing Nanomaterial-Based Optical Spectroscopic Detection of Cancer through Machine Learning

ACS Materials Letters, Год журнала: 2024, Номер 6(10), С. 4697 - 4709

Опубликована: Сен. 18, 2024

Язык: Английский

---