12/24 𝗕𝗲𝗻𝗰𝗵𝗺𝗮𝗿𝗸𝗶𝗻𝗴 𝗣𝗮𝘁𝗵𝗼𝗹𝗼𝗴𝘆 𝗙𝗼𝘂𝗻𝗱𝗮𝘁𝗶𝗼𝗻 𝗠𝗼𝗱𝗲𝗹𝘀 𝗳𝗼𝗿 𝗦𝗽𝗮𝘁𝗶𝗮𝗹 𝗗𝗼𝗺𝗮𝗶𝗻 𝗨𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱𝗶𝗻𝗴 This paper introduces SpaPath-Bench, a representation-level benchmark designed to diagnose spatial representation capability in Pathology Foundation Models (PFMs). It formulates spatial domain identification (SDI) on 42 public paired whole slide image and spatial transcriptomics data, evaluating 19 encoders and seven SDI methods using three complementary criteria. Across 83K runs, SpaPath-Bench reveals different pretraining paradigms capture distinct aspects of tissue spatial architecture, guiding the development of spatially aware computational pathology models. Code and data pipelines are available at bokai-zhao.github.io/SpaPath…. #SpaPathBench #PathologyAI #ComputationalPathology #PFMs #SpatialBiology #RepresentationLearning #MedicalBenchmarks Paper Link: arxiv.org/abs/2605.25764