A novel automated protocol for ice crystal segmentation analysis using Cellpose and Fiji

Abstract

Accurate measurement of ice crystal size is an essential step in quantitative ice recrystallization inhibition (IRI) analysis using the sucrose sandwiching assay (SSA) and splat assay (SA). Here, we introduce a novel method of measuring ice crystal size and shape using Fiji and Cellpose, an anatomical segmentation algorithm, to address the time-consuming and limited number of ice particle determination associated with the mean largest grain size measurement. This new automated approach, displaying rapid segmentation of ∼70 s per image, measures every ice crystal in an image field of view, consequently reducing bias introduced by subjectively selecting the largest crystals in an image. Consistent in determining a diverse set of crystal sizes and shapes, this method allows for the evaluation of ice crystals using Feret's diameter, a parameter that better accounts for irregular particle shape. This method provides new outputs such as standard deviation, particle size distributions of a population of ice crystals, and circularity to characterize and further provide insight into an analyte's IRI ability. Applicable to the SSA, the “shape descriptor” measurement can be used to quantify ice binding. This work presents a novel and accurate approach for ice crystal quantitative analysis.