Artificial intelligence and radiomics in glioma: A bibliometric analysis of key trends and future frontiers
Shichao Liu, Risheng LiangObjective
Gliomas, the most common primary malignant brain tumors, present diagnostic and therapeutic challenges. Artificial intelligence (AI) and radiomics offer promising solutions, yet a macroscopic overview of this rapidly evolving field is lacking. This study provides a bibliometric analysis of research trends, hotspots, and future frontiers to guide clinical application.
Methods
We retrieved literature from five major databases (WoSCC, PubMed, Embase, Scopus, and Cochrane Library) from 2004 to 2025. Using CiteSpace and RStudio, we analyzed publication trends, keyword co-occurrence, collaboration networks, and citation patterns to visualize the research landscape.
Results
A total of 595 publications were analyzed, revealing an exponential growth trend, with 76.3% of articles published between 2020 and 2024. China (191 articles) and the United States (163 articles) were the leading countries in publication volume. The analysis identified deep learning (DL) and radiogenomics as major research hotspots, with a clear paradigm shift from traditional feature engineering to end-to-end automated models.
Conclusion
This bibliometric analysis highlights the rapid evolution of AI and radiomics in glioma research, marked by a paradigm shift from feature-based methods to DL and multi-omics frameworks. Despite significant progress, critical hurdles to clinical translation remain; for example, adherence to the Image Biomarker Standardisation Initiative (IBSI) guidelines is limited, and large-scale, multi-center validation studies are urgently needed. This study provides a crucial roadmap emphasizing the need for explainable AI and robust validation to bridge the gap between research and clinical practice.