Introduction. This study aims to examine how artificial intelligence (AI) supports radiologists in improving workflow efficiency and diagnostic accuracy, focusing on its application in routine imaging tasks and its broader clinical impact. The objective of this study is to explore recent developments in artificial intelligence (AI) applications for musculoskeletal (MSK) magnetic resonance imaging (MRI), with a focus on improving diagnostic accuracy and efficiency, particularly in identifying conditions such as anterior cruciate ligament (ACL) tears and osteoarthritis.
Materials and methods. AI has been increasingly integrated into radiology practice to assist with tasks such as image interpretation, segmentation, and anomaly detection. AI systems utilize deep learning algorithms, which analyze medical images and flag abnormalities, such as fractures, tumors, and vascular conditions. By reducing time spent on repetitive tasks, AI allows radiologists to focus on complex cases. Various studies have evaluated AI's role in detecting critical conditions across imaging modalities such as CT, MRI, and X-ray. Over the last four years, convolutional neural networks (CNNs) and deep learning (DL) models have been at the forefront of MSK imaging research. These models have been trained on large datasets to identify and classify MSK conditions from MRI images. Research has focused on their ability to replicate and potentially surpass the diagnostic performance of human radiologists. This review examines key studies in this field, comparing AI model accuracy with that of human experts.
Results and discussions. AI-assisted diagnostic tools significantly improve efficiency by reducing the time required for interpreting routine imaging. For example, AI systems were found to reduce the time to diagnose conditions by up to 40%, while maintaining or even exceeding human-level diagnostic accuracy in identifying pathologies like lung nodules, fractures, and brain hemorrhages. Radiologists using AI tools reported improved confidence in diagnosis, particularly in challenging or ambiguous cases, while novice radiologists benefited most from AI support due to increased diagnostic consistency. AI algorithms demonstrated high accuracy in diagnosing ACL tears, with CNNs achieving sensitivities of 93-99% and specificities of 87-100%, depending on the homogeneity of the test data. Predictive models for osteoarthritis, integrating clinical biomarkers and MRI data, have shown promise in identifying early stages of disease progression. AI-based bone age assessments are also becoming more reliable, though these applications remain less common in MSK MRI compared to radiographs.
Conclusions. AI plays a crucial role in augmenting radiologists by automating routine tasks and improving diagnostic accuracy, leading to enhanced workflow efficiency. This collaboration between human expertise and AI enables radiologists to deliver faster and more reliable diagnoses, with potential to significantly impact patient outcomes across various clinical settings.
The primary limitations of AI in radiology include variability in performance across different imaging systems and patient demographics. Further research is needed to address generalizability issues and evaluate long-term clinical and cost-effectiveness AI technologies in MSK MRI offer significant potential for enhancing diagnostic accuracy, streamlining clinical workflows, and reducing radiologists’ workloads. These advances could lead to earlier and more accurate diagnosis of MSK conditions such as ACL tears and osteoarthritis.
Despite the promising results, AI models still face challenges related to generalizability across diverse datasets and institutions. External validation and multi-center trials are needed to ensure broader clinical adoption.
List of links:
1. Gitto, S., Serpi, F., Albano, D. et al. AI applications in musculoskeletal imaging: a narrative review. Eur Radiol Exp 8, 22 (2024). https://doi.org/10.1186/s41747-024-00422-8
2. Yi Xian Cassandra Yang et al. “An artificial intelligence boost to MRI lumbar spine reporting.” European Journal of Radiology. July 16, 2024.
3. Fritz, B., Fritz, J. Artificial intelligence for MRI diagnosis of joints: a scoping review of the current state-of-the-art of deep learning-based approaches. Skeletal Radiol 51, 315–329 (2022). https://doi.org/10.1007/s00256-021-03830-8
3. https://www.rheumatology-uoc.gr/images/3o-therino-sxoleio-aktinologias-myoskeletikou/klontzas2-2020.pdf
|
Голосування 
Ця робота ліцензується відповідно до Creative Commons Attribution 4.0 International License
Знайшли помилку? Виділіть помилковий текст мишкою і натисніть Ctrl + Enter