Noninvasive tomography of tumor progression via dynamical changes of T-cells in surrounding microenvironment

Tumor progression is shaped by a continuous competition between malignant expansion and immune access within the surrounding microenvironment. Although spatial assays can reveal where T cells accumulate, most current approaches remain descriptive, treating immune organization as a static pattern and failing to infer the latent resistance structure that governs progression. Here we show that changes in peri-tumoral T-cell distribution can support noninvasive tomography of tumor progression. Using spatially resolved human breast tumor sections and controlled numerical experiments, we formulate immune-tumor interaction within a mean-field game framework in which T cells act as agents responding to a heterogeneous microenvironment while the tumor shapes an equivalent barrier field that restricts infiltration. This framework converts boundary enrichment, layered blocking and core penetration patterns into a constrained inference problem whose solution yields quantitative maps of immune exclusion, barrier strength and progression-associated internal states. Forward simulations show that barrier geometry reshapes immune-density equilibria, and inference experiments show that closely related barrier fields can be recovered from immune-density observations under controlled and tissue-shaped proxy settings. By linking observable T-cell redistribution to latent microenvironmental resistance, this approach establishes a route toward noninvasive, repeatable assessment of tumor progression without direct biopsy of the underlying barrier structure.