Tuesday 08 April 2025
Medical imaging is a crucial aspect of healthcare, providing doctors with vital information to diagnose and treat patients. However, the process of reviewing these images can be time-consuming and prone to human error. A new study has shed light on how artificial intelligence (AI) can help streamline this process, making it faster, more accurate, and more reliable.
Researchers have developed a standardized dataset and evaluation framework for medical imaging quality control tasks, specifically focusing on chest X-rays and computed tomography (CT) reports. This dataset, comprising 161 CXR radiographs and 219 CT reports, is designed to assess the performance of large language models in detecting errors and inconsistencies in these images.
The study evaluated several AI models, including Gemini 2.0-Flash, GPT-4o, InternVL2-8B, QVQ-72B Preview, and Qwen2.5- VL-72B-Instruct. These models were tested on their ability to detect technical errors, such as off-centered projections, malpositioning, and artifacts, in CXR radiographs. The results showed that Gemini 2.0-Flash achieved a remarkable Macro F1 score of 90, indicating its superior generalization ability across diverse quality control categories.
In contrast, the performance of AI models varied significantly when it came to detecting errors in CT reports. DeepSeek-R1 emerged as the top performer, achieving a recall rate of 62.23% and outperforming other models in precision and F1 score. The study highlighted the importance of maintaining model complexity and domain-specific knowledge in medical imaging quality control tasks.
The findings of this study have significant implications for the future of medical imaging. By leveraging AI-powered tools, radiologists can focus on more complex and high-value tasks, such as analyzing image data and providing expert opinions. Additionally, AI-assisted quality control can help reduce errors and improve patient outcomes, ultimately enhancing the overall efficiency and effectiveness of healthcare services.
The study’s limitations are acknowledged by the researchers, who recognize that the dataset is limited to a single institution and primarily consists of Chinese-language reports. Future studies should aim to expand this dataset to include more diverse languages and institutions, as well as explore the potential applications of AI in other medical imaging modalities, such as magnetic resonance imaging (MRI) and ultrasound.
As the field of medical imaging continues to evolve, it is clear that AI has the potential to revolutionize the way we approach quality control.
Cite this article: “Unlocking Medical Imaging Quality Control with Multimodal Large Language Models: A New Era in Radiology Reporting”, The Science Archive, 2025.
Medical Imaging, Artificial Intelligence, Quality Control, Chest X-Rays, Computed Tomography, Large Language Models, Error Detection, Radiographs, Ct Reports, Deep Learning
