“…However, the manual selection of cells of interest and use of commercial software algorithms (e.g., principal component analysis [21] and spectral difference analysis [20]) remain the most common approach to comparative analysis of the spectral patterns. Recently, machine learning methods (e.g., CNNs in combination with U‐Nets [23] and fuzzy learning classification [36]) have been applied to automate cell contour recognition, segmentation, and classification. However, these machine learning techniques still require manual pixel labeling.…”