The Role of Artificial Intelligence in Improving Diagnostic Accuracy in Medical Imaging: A Review
Open Access
- 1 January 2025
- journal article
- review article
- Published by Tech Science Press in Computers, Materials & Continua
- Vol. 85 (2) , 2443-2486
Abstract
This review comprehensively analyzes advancements in artificial intelligence, particularly machine learning and deep learning, in medical imaging, focusing on their transformative role in enhancing diagnostic accuracy. Our in-depth analysis of 138 selected studies reveals that artificial intelligence (AI) algorithms frequently achieve diagnostic performance comparable to, and often surpassing, that of human experts, excelling in complex pattern recognition. Key findings include earlier detection of conditions like skin cancer and diabetic retinopathy, alongside radiologist-level performance for pneumonia detection on chest X-rays. These technologies profoundly transform imaging by significantly improving processes in classification, segmentation, and sequential analysis across diverse modalities such as X-rays, Computed Tomography (CT), Magnetic Resonance Imaging (MRI), and ultrasound. Specific advancements with Convolutional Neural Networks, Recurrent Neural Networks, and ensemble learning techniques have facilitated more precise diagnosis, prediction, and therapy planning. Notably, Generative Adversarial Networks address limited data through augmentation, while transfer learning efficiently adapts models for scarce labeled datasets, and Reinforcement Learning shows promise in optimizing treatment protocols, collectively advancing patient care. Methodologically, a systematic review (2015–2024) used Scopus and Web of Science databases, yielding 7982 initial records. Of these, 1189 underwent bibliometric analysis using the R package ‘Bibliometrix’, and 138 were comprehensively reviewed for specific findings. Research output surged over the decade, led by Institute of Electrical and Electronics Engineers (IEEE) Access (19.1%). China dominates publication volume (36.1%), while the United States of America (USA) leads total citations (5605), and Hong Kong exhibits the highest average (55.60). Challenges include rigorous validation, regulatory clarity, and fostering clinician trust. This study highlights significant emerging trends and crucial future research directions for successful AI implementation in healthcare.Keywords
This publication has 128 references indexed in Scilit:
- Impacts of the advancement in artificial intelligence on laboratory medicine in low‐ and middle‐income countries: Challenges and recommendations—A literature reviewHealth Science Reports, 2024
- Brain tumor segmentation using K‐means clustering and deep learning with synthetic data augmentation for classificationMicroscopy Research and Technique, 2021
- Deep Learning—A Technology With the Potential to Transform Health CareJAMA, 2018
- Medical Image Synthesis for Data Augmentation and Anonymization Using Generative Adversarial Networks2018
- Development and Validation of a Deep Learning System for Diabetic Retinopathy and Related Eye Diseases Using Retinal Images From Multiethnic Populations With DiabetesJAMA, 2017
- Deep reinforcement learning for automated radiation adaptation in lung cancerMedical Physics, 2017
- Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus PhotographsJAMA, 2016
- Co‐citation analysis, bibliographic coupling, and direct citation: Which citation approach represents the research front most accurately?Journal of the American Society for Information Science and Technology, 2010
- Overview of Semi-Supervised Learning2009
- What do we know about the h index?Journal of the American Society for Information Science and Technology, 2007