Radiomics and deep learning in liver diseases.

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence

Background:

• Radiomics and deep learning are emerging computerized image analysis methods.
• These techniques extract high-dimensional features for diagnostic and predictive tasks in medical imaging.
• Their potential extends beyond visual analysis in liver imaging.

Purpose of the Study:

• To review the technical aspects of radiomics and deep learning.
• To summarize recent studies applying these techniques in liver radiology.
• To highlight their potential in clinical practice.

Main Methods:

• Review of recent scientific literature on radiomics and deep learning in liver radiology.
• Analysis of studies focusing on technical feasibility and preliminary applications.
• Discussion of various applications including fibrosis staging, tumor prognostication, and segmentation.

Main Results:

• Radiomics and deep learning show promise in liver fibrosis staging.
• These methods aid in prognostication of malignant liver tumors.
• Automated detection, characterization, segmentation, and body composition analysis are potential applications.

Conclusions:

• Radiomics and deep learning have demonstrated potential in diverse liver imaging applications.
• Current studies are largely preliminary, focusing on technical feasibility.
• Extensive clinical validation is essential for routine integration into liver radiology practice.