Supporting Data for Systematic Review on Transfer Learning for Breast Mammography

This extraction table (Excel) contains one row per included study with fields for Title/Author, dataset(s), preprocessing, pretrained model(s), transfer learning approach (feature extraction vs fine‑tuning), reported metrics (accuracy, AUC, sensitivity, specificity, precision, recall, F1), code availability. It supports reproducible synthesis and can be reused for meta‑analyses and benchmarking.

This PRISMA flow diagram summarizes the study selection process: records identified across PubMed, Scopus, IEEE Xplore, and SpringerLink; duplicates removed; title/abstract screening; full‑text assessment; and final inclusion of 154 studies.

本提取表（Excel格式）为每一篇纳入研究单独设置一行，包含以下字段：标题/作者、数据集（dataset）、预处理流程、预训练模型（pretrained model）、迁移学习（transfer learning）方法（特征提取（feature extraction）与微调（fine-tuning）二选一）、报告的评价指标（准确率、AUC、灵敏度、特异度、精确率、召回率、F1值）以及代码可用性。本提取表可支撑可复现的研究整合，亦可用于元分析（meta-analysis）与基准评测（benchmarking）。
本PRISMA流程图（PRISMA flow diagram）总结了文献筛选全流程：跨PubMed、Scopus、IEEE Xplore及SpringerLink数据库检索得到的文献记录、去除重复文献、标题与摘要筛查、全文评估，最终纳入154项研究。