Spatial feature analysis for digital pathology images

Abstract

Systems and methods relate to processing digital pathology images. More specifically, depictions of objects of a first class (e.g., lymphocytes) and depictions of objects of a second class (e.g., tumor cells) are detected. Locations of each biological object depiction are identified, which are used to generate multiple spatial-distribution metrics that characterize where depictions of objects of a first class are located relative to objects of a second class. The spatial-distribution metrics are used to generate a result corresponding to a predicted biological state of or a potential treatment of a subject. For example, the result may predict whether and/or an extent to which lymphocytes have infiltrated a tumor, whether checkpoint blockade therapy would be an effective treatment for the subject, and/or whether a subject is eligible for a clinical trial.