
Motivation & Challenge: Deep learning methods for radiology report generation (RRG) grew fast, but the literature lacked a unified map.
As first author, I proposed a three-way taxonomy—visual-only, textual-only, and cross-modal approaches—covering benchmarks (IU X-Ray, MIMIC-CXR, CX-CHR, COV-CTR) and metrics.
Results & Venues: A structured taxonomy, benchmark reference, and open-challenge analysis. Companion GitHub repository curates 100+ papers, references, datasets, and toolkits. Submitted to IEEE TPAMI.
Link: https://arxiv.org/pdf/2311.14199
GitHub: https://github.com/synlp/RRG-Review
1 like · 1 view · Sep 15, 2026
Motivation & Challenge: Deep learning methods for radiology report generation (RRG) grew fast, but the literature lacked a unified map.