Developing Approaches to Incorporate Donor Lung CT Images into Machine Learning Models to Predict Severe Primary Graft Dysfunction after Lung Transplantation

American Journal of Transplantation, Год журнала: 2025, Номер unknown

Опубликована: Фев. 1, 2025

Primary graft dysfunction (PGD) is a common complication after lung transplantation associated with poor outcomes. Although risk factors have been identified, the complex interactions between clinical variables affecting PGD are not well understood, which can complicate decisions about donor acceptance. Previously, we developed machine learning (ML) model to predict grade 3 using and recipient electronic health record (EHR) data, but it lacked granular information from CT scans, routinely assessed during offer review. In this study, used gated approach determine optimal methods for analyzing scans among patients receiving first-time, bilateral transplants at single center over 10 years. We four computer vision approaches fused best EHR data three points in ML process. A total of 160 had donor-lung analysis. The imaging-only employed 3D ResNet model, yielding median (IQR) AUROC AUPRC 0.63 (0.49 - 0.72) 0.48 (0.35 0.6), respectively. Combining imaging late fusion provided highest performance, 0.74 (0.59 0.85) 0.61 (0.47 0.72),

Язык: Английский

---

Advancing ultrasonography with AI: recent innovations and future perspectives

Holistic Integrative Oncology, Год журнала: 2025, Номер 4(1)

Опубликована: Апрель 8, 2025

Язык: Английский

---

Enhanced breast cancer diagnosis using modified InceptionNet-V3: a deep learning approach for ultrasound image classification

Frontiers in Physiology, Год журнала: 2025, Номер 16

Опубликована: Апрель 22, 2025

Breast cancer (BC) is a malignant neoplasm that originates in the mammary gland's cellular structures and remains one of most prevalent cancers among women, ranking second cancer-related mortality after lung cancer. Early accurate diagnosis crucial due to heterogeneous nature breast its rapid progression. However, manual detection classification are often time-consuming prone errors, necessitating development automated reliable diagnostic approaches. Recent advancements deep learning have significantly improved medical image analysis, demonstrating superior predictive performance using ultrasound images. Despite these advancements, training models from scratch can be computationally expensive data-intensive. Transfer learning, leveraging pre-trained on large-scale datasets, offers an effective solution mitigate challenges. In this study, we investigate compare multiple deep-learning for transfer learning. The evaluated architectures include modified InceptionV3, GoogLeNet, ShuffleNet, AlexNet, VGG-16, SqueezeNet. Additionally, propose neural network model integrates features InceptionV3 further enhance performance. experimental results demonstrate achieves highest accuracy 99.10%, with recall 98.90%, precision 99.00%, F1-score 98.80%, outperforming all other given datasets. achieved findings underscore potential proposed approach enhancing confirm superiority tasks.

Язык: Английский

---

Smart Plant Disease Diagnosis Using Multiple Deep Learning and Web Application Integration

Journal of Agriculture and Food Research, Год журнала: 2025, Номер 21, С. 101948 - 101948

Опубликована: Апрель 23, 2025

Язык: Английский

---

Deep learning approaches for classification tasks in medical X-ray, MRI, and ultrasound images: a scoping review

BMC Medical Imaging, Год журнала: 2025, Номер 25(1)

Опубликована: Май 7, 2025

Язык: Английский

---

Impact of test set composition on AI performance in pediatric wrist fracture detection in X-rays

European Radiology, Год журнала: 2025, Номер unknown

Опубликована: Май 16, 2025

To evaluate how different test set sampling strategies-random selection and balanced sampling-affect the performance of artificial intelligence (AI) models in pediatric wrist fracture detection using radiographs, aiming to highlight need for standardization design. This retrospective study utilized open-sourced GRAZPEDWRI-DX dataset 6091 radiographs. Two sets, each containing 4588 images, were constructed: one a approach based on case difficulty, projection type, presence other random selection. EfficientNet YOLOv11 trained validated 18,762 radiographs tested both sets. Binary classification object tasks evaluated metrics such as precision, recall, F1 score, AP50, AP50-95. Statistical comparisons between sets performed nonparametric tests. Performance significantly decreased with more challenging cases. For example, precision from 0.95 0.83 set. Similar trends observed accuracy, indicating that easy-to-recognize cases poorly complex ones. These results consistent across all model variants tested. AI exhibit reduced when datasets difficult cases, compared randomly selected highlights importance constructing representative standardized account clinical complexity ensure robust real-world settings. Question Do strategies samples' have an influence deep learning models' detection? Findings drops Clinical relevance Without reflect complexities, may be overestimated, limiting utility

Язык: Английский

---

Automated CT segmentation for lower extremity tissues in lymphedema evaluation using deep learning

European Radiology, Год журнала: 2025, Номер unknown

Опубликована: Май 16, 2025

Язык: Английский

---

Improving Breast Cancer Diagnosis in Ultrasound Images Using Deep Learning with Feature Fusion and Attention Mechanism

Academic Radiology, Год журнала: 2025, Номер unknown

Опубликована: Май 1, 2025

Язык: Английский

---