DKI-based spatial heterogeneity analysis for tumor-infiltrating lymphocytes level in breast cancer patients

Motivation: Verification that DKI can accurately yet noninvasively detect intratumoral complexity, water diffusion, and lesion heterogeneity information, without contrast agents or biopsy, would encourage wider application of this technique. Goal(s): To ascertain the performance of the TILs level prediction model. Approach: A DKI-based tumor survival environment model was developed and used for predicting TILs level in breast cancer patients. Tumors segmented in microscopic habitat clusters were explored for heterogeneity with DKI data. Results: First order features extracted from tumor clusters based on DKI data can properly predict TILs level. Impact: The stable predictive performance of DKI/TILs can help clinicians make more accurate diagnoses; commercialization is expected.