Machine Learning Models Using Spectral CT Data Help Better Classify Lung Nodules as Benign or Malignant
-
Lung cancer often presents as solitary pulmonary nodules (SPNs). Accurately identifying SPNs as benign or malignant is important to guide treatment.
-
Dual-layer detector spectral CT (DLCT) provides quantitative imaging parameters that may help distinguish benign and malignant SPNs.
-
This study evaluated DLCT parameters and clinical data from 250 SPN patients, and built machine learning models to classify SPNs.
-
The logistic regression model combining DLCT parameters and clinical data performed best, with accuracy over 80% for identifying benign vs malignant SPNs.
-
DLCT imaging features quantified by machine learning models could improve differentiation of benign and malignant SPNs to guide clinical management.