ChatGPT-4 demonstrates near-perfect accuracy in classifying pancreatic cysts on MRI and CT, matching radiologist performance.
Key Details
- 1Study from Memorial Sloan Kettering used ChatGPT-4 to evaluate 3,198 MRI and CT scans of 991 adults under surveillance for pancreatic cysts.
- 2ChatGPT-4 was assessed on its ability to identify nine variables crucial for monitoring cyst progression.
- 3LLM accuracy for categorical variables ranged from 97% (solid component lesions) to 99% (calcific lesions).
- 4Accuracy for continuous variables ranged from 92% (cyst size) to 97% (main pancreatic duct size).
- 5ChatGPT-4's performance was found to be equivalent to manual radiologist chart review, which is the clinical gold standard.
- 6Authors note limitation: only one AI model was tested; further research is needed.
Why It Matters
This study suggests that advanced language models like ChatGPT-4 can automate time-consuming radiological classification tasks with expert-level accuracy, potentially streamlining cyst surveillance workflows and augmenting clinical decision support.
Source
AuntMinnie
Related News
•HealthExec
AI's Expanding Role in Healthcare and Implications for Radiology
A series of thought leaders and institutions weigh in on AI's transformative potential in healthcare, with emphasis on radiology adoption and responsible use.
•Radiology Business
Most Radiology AI Users Lack Clear Evidence of Financial ROI
Survey finds over 75% of radiology organizations using AI lack clear, quantified ROI data.
•AuntMinnie
Toronto Study: LLMs Must Cite Sources for Radiology Decision Support
University of Toronto researchers found that large language models (LLMs) such as DeepSeek V3 and GPT-4o offer promising support for radiology decision-making in pancreatic cancer when their recommendations cite guideline sources.