Using CT Scans and Deep Learning to Predict Pancreatic Infections
Author Information
Author(s): Huang Bingyao, Gao Yi, Wu Lina
Primary Institution: Shengjing Hospital of China Medical University
Hypothesis
Can body composition assessed through non-contrast CT scans predict the risk of infectious pancreatic necrosis in patients with acute necrotizing pancreatitis?
Conclusion
The study found that specific body composition features, such as the distribution of fat and muscle, are significant risk factors for developing infectious pancreatic necrosis.
Supporting Evidence
- Skewness and uniformity of visceral adipose tissue were significant predictors of infectious pancreatic necrosis.
- Subcutaneous adipose tissue volume was larger in patients who developed infectious pancreatic necrosis.
- Abdominal muscle density distribution was more heterogeneous in patients with infectious pancreatic necrosis.
Takeaway
Doctors can use special scans to see how much fat and muscle people have, which helps them figure out if someone might get a serious infection in their pancreas.
Methodology
The study analyzed CT scans of 315 patients with acute necrotizing pancreatitis, using deep learning to segment body tissues and extract radiomic features.
Potential Biases
Potential bias due to the retrospective nature of the study and reliance on CT imaging without contrast.
Limitations
The study is limited by its single-center design and the lack of detailed analysis on patient prognosis.
Participant Demographics
Patients diagnosed with acute necrotizing pancreatitis, with a mean age of approximately 47.91 years.
Statistical Information
P-Value
p=0.004 for skewness of VAT
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website