| Abstract | The reliable measurement of cellulose nanocrystal (CNC) particle morphology is vital for ongoing CNC production optimization and application development. A semi-automatic image analysis program, SMART-AFM, was developed for CNC particle size measurements from atomic force microscopy (AFM) images and subsequently used to analyze the AFM images acquired from four laboratories that participated in the interlaboratory comparison (ILC) study by Bushell et al.. SMART-AFM demonstrated a 40 times faster analysis compared to the manual approach used in the ILC. A detailed image-to-image analysis showed that SMART-AFM produced similarly accurate results as compared to the manual method for CNC identification, as well as length and height measurements. SMART-AFM “correctly” identified CNCs 82–90% and manual analysis identified 83–94%, depending on the given laboratory image dataset. While SMART-AFM reported mean length values consistently 7–16.5% lower than the manual approach—attributed to trimming CNC tails during segmentation—CNC height measurements were in closer agreement—an average difference of 4.2%. Moreover, SMART-AFM CNC identification and measurement demonstrated lower variability across images from different laboratories, potentially indicating higher identification consistency compared to the subjective and qualitative nature of manual analysis. The strengths and challenges of SMART-AFM in analyzing images affected by AFM imaging artifacts is systematically studied. An analysis is presented to reaffirm the importance of measuring 300–500 CNCs to ensure a representative, reliable measurement results. Overall, SMART-AFM is established as a standardized CNC identification and measurement tool, with improved speed compared to the manual alternative, and proven consistency and reliability. The SMART-AFM code is publicly available in Github™. |
|---|
