Mammo-CLIP：一种提升乳腺X线摄影数据效率与鲁棒性的视觉语言基础模型
Mammo-CLIP: A Vision Language Foundation Model to Enhance Data Efficiency and Robustness in Mammography
AI总结：
针对乳腺癌CAD训练数据不足问题，提出首个基于乳腺X线摄影-报告对的视觉语言基础模型Mammo-CLIP，并引入特征归因方法Mammo-FActOR，在公开数据集上验证了其数据效率与鲁棒性。
摘要：
在乳腺癌检测的计算机辅助诊断（CAD）中，缺乏大规模且多样化的训练数据一直是阻碍该系统应用的问题之一。近期，通过视觉语言模型（VLM，如CLIP）在大规模图像-文本数据集上进行预训练，部分解决了计算机视觉（CV）中鲁棒性和数据效率的问题。本文提出了Mammo-CLIP，这是首个在大量筛查性乳腺X线摄影-报告对上进行预训练的VLM，旨在解决数据集多样性和规模方面的挑战。我们在两个公开数据集上的实验表明，该模型在分类和定位对乳腺癌检测至关重要的各种乳腺X线摄影属性方面表现出色，展示了与CV中CLIP类似的数据效率和鲁棒性。我们还提出了Mammo-FActOR，一种新颖的特征归因方法，以在乳腺X线摄影报告中提供句子级别的表示空间解释。代码已公开：https://github.com/batmanlab/Mammo-CLIP。
英文摘要：
The lack of large and diverse training data on Computer-Aided Diagnosis (CAD) in breast cancer detection has been one of the concerns that impedes the adoption of the system. Recently, pre-training with large-scale image text datasets via Vision-Language models (VLM) (\eg CLIP) partially addresses the issue of robustness and data efficiency in computer vision (CV). This paper proposes Mammo-CLIP, the first VLM pre-trained on a substantial amount of screening mammogram-report pairs, addressing the challenges of dataset diversity and size. Our experiments on two public datasets demonstrate strong performance in classifying and localizing various mammographic attributes crucial for breast cancer detection, showcasing data efficiency and robustness similar to CLIP in CV. We also propose Mammo-FActOR, a novel feature attribution method, to provide spatial interpretation of representation with sentence-level granularity within mammography reports. Code is available publicly: \url{https://github.com/batmanlab/Mammo-CLIP}.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
