Patient education strategies in pediatric orthopaedics: using ChatGPT to answer frequently asked questions on scoliosis

Spine Deformity, Journal Year: 2025, Volume and Issue: unknown

Published: April 5, 2025

Language: Английский

---

Evaluating the performance of GPT-3.5, GPT-4, and GPT-4o in the Chinese National Medical Licensing Examination

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 23, 2025

Language: Английский

---

Reliability analysis of automated Cobb angle measurement using artificial intelligence models in scoliosis patients

Medical Biological Science and Engineering, Journal Year: 2025, Volume and Issue: 8(1), P. 14 - 19

Published: Jan. 22, 2025

Language: Английский

---

Comparative Evaluation of Large Language and Multimodal Models in Detecting Spinal Stabilization Systems on X-Ray Images

Journal of Clinical Medicine, Journal Year: 2025, Volume and Issue: 14(10), P. 3282 - 3282

Published: May 8, 2025

Background/Objectives: Open-source AI models are increasingly applied in medical imaging, yet their effectiveness detecting and classifying spinal stabilization systems remains underexplored. This study compares ChatGPT-4o (a large language model) BiomedCLIP multimodal analysis of posturographic X-ray images (AP projection) to assess accuracy identifying the presence, type (growing vs. non-growing), specific system (MCGR PSF). Methods: A dataset 270 (93 without stabilization, 80 with MCGR, 97 PSF) was analyzed manually by neurosurgeons evaluated using a three-stage AI-based questioning approach. Performance assessed via classification accuracy, Gwet’s Agreement Coefficient (AC1) for inter-rater reliability, two-tailed z-test statistical significance (p < 0.05). Results: The results indicate that GPT-4o demonstrates high systems, achieving near-perfect recognition (97–100%) presence or absence stabilization. However, its consistency is reduced when distinguishing complex growing-rod (MCGR) configurations, agreement scores dropping significantly (AC1 = 0.32–0.50). In contrast, displays greater response 1.00) but struggles detailed classification, particularly recognizing PSF (11% accuracy) MCGR (4.16% accuracy). Sensitivity revealed GPT-4o’s superior stability hierarchical tasks, while excelled binary detection showed performance deterioration as complexity increased. Conclusions: These findings highlight robustness clinical AI-assisted diagnostics, differentiation whereas BiomedCLIP’s precision may require further optimization enhance applicability radiographic evaluations.

Language: Английский

---

Synthetic Genitourinary Image Synthesis via Generative Adversarial Networks: Enhancing Artificial Intelligence Diagnostic Precision

Journal of Personalized Medicine, Journal Year: 2024, Volume and Issue: 14(7), P. 703 - 703

Published: June 30, 2024

In the realm of computational pathology, scarcity and restricted diversity genitourinary (GU) tissue datasets pose significant challenges for training robust diagnostic models. This study explores potential Generative Adversarial Networks (GANs) to mitigate these limitations by generating high-quality synthetic images rare or underrepresented GU tissues. We hypothesized that augmenting data pathology models with GAN-generated images, validated through pathologist evaluation quantitative similarity measures, would significantly enhance model performance in tasks such as classification, segmentation, disease detection.

Language: Английский

---

Synthetic Genitourinary Image Synthesis via Generative Adversarial Networks: Enhancing AI Diagnostic Precision

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: May 21, 2024

Abstract In the realm of computational pathology, scarcity and restricted diversity genitourinary (GU) tissue datasets pose significant challenges for training robust diagnostic models. This study explores potential Generative Adversarial Networks (GANs) to mitigate these limitations by generating high-quality synthetic images rare or underrepresented GU tissues. We hypothesized that augmenting data pathology models with GAN-generated images, validated through pathologist evaluation quantitative similarity measures, would significantly enhance model performance in tasks such as classification, segmentation, disease detection. To test this hypothesis, we employed a GAN produce eight different The quality was rigorously assessed using Relative Inception Score (RIS) 17.2 ± 0.15 Fréchet Distance (FID) stabilized at 120, metrics reflect visual statistical fidelity generated real histopathological images. Additionally, received an 80% approval rating from board-certified pathologists, further validating their realism utility. used alternative Spatial Heterogeneous Recurrence Quantification Analysis (SHRQA) assess prostate tissue. allowed us make comparison between original context features, which were pathologist’s evaluation. Future work will focus on implementing deep learning evaluate augmented provide more comprehensive understanding utility enhancing workflows. not only confirms feasibility GANs augmentation medical image analysis but also highlights critical role addressing dataset imbalance. refining generative even diverse complex representations, potentially transforming landscape diagnostics AI-driven solutions. CONSENT FOR PUBLICATION All authors have provided consent publication.

Language: Английский

---

A generative adversarial network to Reinhard stain normalization for histopathology image analysis

Ain Shams Engineering Journal, Journal Year: 2024, Volume and Issue: 15(10), P. 102955 - 102955

Published: July 14, 2024

Histopathology image analysis is paramount importance for accurate diagnosing diseases and gaining insight into tissue properties. The significant challenge of staining variability continues. This research work presents a new method that merges deep learning with Reinhardstain normalization, aiming to revolutionize histopathology analysis. multi-data stream attention-based generative adversarial network an innovative architecture designed enhance histopathological by integrating multiple data streams, attention mechanisms, networks improved feature extraction quality. Multi-data capitalizes on mechanisms process multi-modal efficiently, enhancing ensuring robust performance even in the presence variations. approach excels exact disease detection classification, emerging as invaluable tool both clinical diagnoses endeavors across diverse datasets. obtained accuracy proposed SCAN dataset 97.75%, BACH 99.50% Break His 99.66%. significantly advances analysis, offering diagnostic deeper insights networks. enhances extraction, quality, overall effectiveness medical

Language: Английский

---

Evaluation of Mis-Selection of End Vertebrae and Its Effect on Measuring Cobb Angle and Curve Length in Adolescent Idiopathic Scoliosis

Journal of Clinical Medicine, Journal Year: 2024, Volume and Issue: 13(15), P. 4562 - 4562

Published: Aug. 5, 2024

Background: The Cobb angle is critical in assessing adolescent idiopathic scoliosis (AIS) patients. This study aimed to evaluate the error selecting upper- and lower-end vertebrae on AIS digital X-rays by experienced novice observers its correlation with measuring determining length of scoliotic curves. Methods: Using TraumaMeter v.873 software, eight raters independently evaluated 68 Results: percentage upper-end vertebra selection was higher than for (44.7%, CI95% 41.05–48.3 compared 35%, 29.7–40.4). mean bias (MBE) 0.45 (CI95% 0.38–0.52) 0.35 (CI% 0.69–0.91) vertebra. errors choice end lower novices. There a positive (r = 0.673, p 0.000) between Conclusions: We can conclude that are common among observers, greater frequency vertebrae. Contrary consensus, accuracy curve limited method’s reliance correct

Language: Английский

---