Radiomics-Based Machine Learning in the Diagnosis of Type-B Aortic Dissection on Computed Tomography Images.

Area of Science:

• Radiology
• Artificial Intelligence
• Medical Imaging Analysis

Background:

• Type-B aortic dissection (TBAD) is a critical cardiovascular condition.
• Accurate and timely diagnosis of TBAD is essential for patient outcomes.
• Computed tomography (CT) is a primary imaging modality for aortic diseases.

Purpose of the Study:

• To evaluate the diagnostic performance of a radiomics-based machine learning model for detecting TBAD.
• To assess the value of radiomics features extracted from CT images in identifying TBAD.
• To compare the predictive accuracy of different machine learning models in TBAD detection.

Main Methods:

• Retrospective analysis of 200 patients (100 TBAD, 100 non-TBAD).
• Extraction of radiomics features from non-contrast CT images.
• Dimensionality reduction and model construction using LASSO, with evaluation via ROC curves.

Main Results:

• Machine learning models demonstrated high prediction accuracy (AUC > 0.9) in the validation set.
• A nomogram model achieved the highest AUC values (0.991 training, 0.998 validation).
• The nomogram exhibited superior calibration and clinical utility in both cohorts.

Conclusions:

• Radiomics analysis combined with machine learning provides a robust method for TBAD detection.
• The developed predictive model shows significant potential for improving TBAD diagnosis accuracy.
• CT-based radiomics offers a valuable tool for identifying Type-B aortic dissection.