Texture Features of Computed Tomography Image under the Artificial Intelligence Algorithm and Its Predictive Value...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Colorectal liver metastases (CRLM) are a significant challenge in colorectal cancer management.
• Early and accurate detection of CRLM is crucial for effective treatment and improved patient outcomes.

Purpose of the Study:

• To investigate the predictive capability of artificial intelligence (AI) algorithms using computed tomography (CT) image texture features for colorectal liver metastases (CRLM).
• To evaluate the performance of various AI classifiers in predicting CRLM and identify the most effective model.

Main Methods:

• Analysis of CT image texture features from 150 colorectal cancer patients categorized into three groups (A: initial CRLM, B: follow-up CRLM, C: no CRLM).
• Application of six AI classifiers to predict CRLM, with performance metrics including prediction accuracy, sensitivity, and specificity.
• Utilized receiver operator characteristic (ROC) curves to assess classifier performance.

Main Results:

• The logistic regression (LR) classifier exhibited the highest prediction accuracy, sensitivity, and specificity among all evaluated classifiers.
• The LR classifier showed superior prediction performance in patients with early metastasis (group B1) and later metastasis (group B3) compared to intermediate metastasis (group B2).
• AI-based texture feature analysis of CT images demonstrated a significant predictive effect for CRLM.

Conclusions:

• AI algorithms analyzing CT image texture features provide a valuable tool for predicting CRLM.
• The logistic regression (LR) classifier is highly effective and clinically valuable for CRLM prediction, supporting its widespread application.
• These findings offer guiding significance for the early diagnosis and treatment strategies of colorectal liver metastases.