发表机构
Faculty of Engineering, University of Peradeniya; School of Computing Technologies, RMIT University; School of Engineering & Digital Technologies, University of Southern Queensland; School of Engineering & Technologies, UNSW（佩拉德尼亚大学工程学院; 皇家墨尔本理工大学计算技术学院; 南昆士兰大学工程与数字技术学院; 新南威尔士大学工程与技术学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出M3-Score，基于RadioDINO-s16的评估框架，通过保真度、记忆和覆盖度三个独立轴全面评估生成式放射学图像模型，在BraTS脑MRI上验证了其有效性和稳定性。
AI 中文摘要
放射学生成模型的定量评估仍然具有挑战性。临床相关的结构通常较小且不常见，从自然图像中学到的特征空间可能无法很好地表示它们，并且单一的汇总分数无法区分有限的保真度与有限的多样性。本研究提出了医学多轴最大均值差异评分（M3-Score），这是一种基于RadioDINO-s16的评估框架，RadioDINO-s16是一个在放射学图像上预训练的冻结视觉变换器。M3-Score在预指定的编码器深度报告三个互补的轴：保真度，通过最终块处的无偏多带宽径向基函数（RBF）MMD²测量；记忆，通过75%深度处的最近邻距离测量；覆盖度，定义为在33%深度处，具有生成邻居在其k最近邻半径内的真实图像的占比。参考集跨受试者进行采样，以限制相关切片的影响。在BraTS脑MRI上，保真度轴对五个严重程度递增的比较集进行了排序（Spearman ρ=1.00），并且一个受试者不相交的真实集产生了MMD²=0（置换p=1）。一个无条件去噪扩散概率模型实现了MMD²=0.073（95%置信区间[0.071,0.080]），但仅覆盖了真实分布的38%。在渐进式模式丢弃下，k-NN流形召回在所有十二个编码器块中增加，而所提出的覆盖度估计器单调减少（ρ=-1.00）。RadioDINO-s16特征将真实脑MRI与生成样本区分开来，ROC-AUC为0.819，而InceptionV3为0.555，CLIP为0.582。在二十倍的样本量范围内，平均M3值变化了1.05倍，而Fréchet Inception Distance则变化了2.52倍。
英文摘要
Quantitative evaluation of generative models for radiology remains challenging. Clinically relevant structures are often small and infrequent, feature spaces learned from natural images may represent them poorly, and a single summary score cannot distinguish limited fidelity from limited diversity. This study proposes the Medical Multi-axis Maximum Mean Discrepancy score (M3-Score), an evaluation framework based on RadioDINO-s16, a frozen vision transformer pretrained on radiology images. M3-Score reports three complementary axes computed at pre-specified encoder depths: \emph{fidelity}, measured by an unbiased multi-bandwidth radial basis function (RBF) MMD$^2$ at the final block; \emph{memorization}, measured by nearest-neighbor distances at 75\% depth; and \emph{coverage}, defined as the fraction of real images with a generated neighbor within their $k$-nearest-neighbor radius at 33\% depth. Reference sets are sampled across subjects to limit the influence of correlated slices. On BraTS brain MRI, the fidelity axis ordered five comparison sets of increasing severity (Spearman $ρ= 1.00$), and a subject-disjoint real set yielded $\mathrm{MMD}^2 = 0$ (permutation $p = 1$). An unconditional denoising diffusion probabilistic model achieved $\mathrm{MMD}^2 = 0.073$ (95\% confidence interval $[0.071, 0.080]$) but covered only 38\% of the real distribution. Under progressive mode dropping, $k$-NN manifold recall increased at all twelve encoder blocks, whereas the proposed coverage estimator decreased monotonically ($ρ= -1.00$). RadioDINO-s16 features separated real brain MRI from generated samples with a ROC-AUC of 0.819, compared with 0.555 for InceptionV3 and 0.582 for CLIP. Across a twentyfold range of sample sizes, the mean M3 value varied by a factor of 1.05, compared with 2.52 for the Fréchet Inception Distance.
