THE CURRENT STATE OF ARTIFICIAL INTELLIGENCE IN RADIOLOGY – A REVIEW OF THE BASIC CONCEPTS, APPLICATIONS, AND CHALLENGES

Introduction: Artificial intelligence (AI) is defined as an artificial entity capable of solving problems, learning from experience, and performing tasks such as pattern recognition and inductive reasoning. In radiology, AI aims to assist with image analysis and interpretation, potentially reducing...

Full description

Saved in:
Bibliographic Details
Main Author: Mariana Yordanova
Format: Article
Language:English
Published: Peytchinski Publishing 2025-03-01
Series:Journal of IMAB
Subjects:
Online Access:https://www.journal-imab-bg.org/issues-2025/issue1/2025vol31-issue1-6103-6107.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Introduction: Artificial intelligence (AI) is defined as an artificial entity capable of solving problems, learning from experience, and performing tasks such as pattern recognition and inductive reasoning. In radiology, AI aims to assist with image analysis and interpretation, potentially reducing human error and alleviating the radiological workload. As machine calculation capacity advances, AI's role in radiomics—extracting numerous quantitative features from medical images—could significantly enhance diagnostic accuracy. Materials and Methods: This review article synthesizes recent advancements in the field by examining studies from the past four years, ensuring the information is current and highlights significant findings and emerging themes. Results and Discussion: Machine learning in radiology focuses on developing algorithms that analyze medical images without explicitly programmed rules, divided into supervised and unsupervised learning. Deep learning, especially deep convolutional neural networks (CNNs), has become a prominent approach, mimicking brain functions to process images through multiple layers. CNNs excel in tasks like lesion detection and disease classification, aiding radiologists in diagnosing conditions more accurately. There are numerous applications of AI in radiology - image segmentation, objective quantification, detecting and highlighting suspicious areas, image processing and optimization, longitudinal analysis, potentially diagnosis recognition, triage, and reporting aid. The adoption of radiological AI faces several challenges, including substantial hardware requirements, data quality and quantity limitations, high false positive rates, the "black box" problem, and the narrow focus of current AI applications, which restricts their clinical usefulness. Conclusion: AI is transforming radiology with diverse applications, but it still needs development to match human expertise.
ISSN:1312-773X