Abstract
The placenta is a vital organ that supports the developing fetus during pregnancy. Histologic examination of the placenta can reveal abnormalities in morphology and structure that impact placental function. Machine learning (ML) models have been successfully developed for digital pathology, leveraging rich image datasets from human tissue. ML models can be advantageous to placenta researchers, either by supplementing pathologist expertise or providing knowledge to inform future hypothesis generation. Research projects fall into several categories: Cell classification methods have been introduced to the placental disc and membranes. Cell classification is useful as a "bottom up" approach to characterizing tissue, using smaller image inputs than at a tissue region or whole slide level. Classification of normal tissues, cells, and development can identify pathologies that deviate. Several studies have identified pathologies within the great obstetric syndromes or placental inflammation. These studies often use mechanisms to aggregate findings from small images patches up to the whole slide level. Digital pathology slides are rich with data that inform our knowledge of placental function and disease - while many articles focus on model design and performance, the features extracted can add valuable biological and clinical knowledge to the field.