Semi-quantitative scoring criteria based on multiple staining methods combined with machine learning to evaluate...

Area of Science:

• Biomaterials Science
• Cell Biology
• Biotechnology

Background:

• Decellularized extracellular matrix (dECM) biomaterials require rigorous quality control.
• Detecting residual nuclei in dECM is crucial for biocompatibility but challenging due to impurities.
• Current methods for residual nuclei detection in dECM lack intelligent analysis and are prone to interference.

Purpose of the Study:

• To develop a more accurate and objective method for detecting residual nuclei in dECM biomaterials.
• To overcome limitations of existing methods in differentiating nuclei from impurities.
• To enhance the efficiency and reliability of dECM quality assessment.

Main Methods:

• Utilized four staining techniques: hematoxylin-eosin, acetocarmine, Feulgen reaction, and 4',6-diamidino-2-phenylindole (DAPI).
• Quantitatively evaluated staining parameters (area, perimeter, grayscale) to create a semi-quantitative scoring system.
• Employed machine learning models trained on nuclear parameter features for automatic nuclei identification.

Main Results:

• No single staining method could reliably distinguish nuclei from impurities.
• A semi-quantitative scoring table achieved over 98% accuracy in identifying individual cell nuclei.
• Combining four staining methods eliminated false positives from impurity contamination.
• The artificial intelligence model achieved over 90% accuracy in detecting residual nuclei.

Conclusions:

• Multidimensional parameter analysis integrated with machine learning significantly enhances residual nuclei detection accuracy in dECM.
• This novel approach offers a reliable, objective, and efficient method for evaluating dECM biomaterials.
• The developed AI model provides a robust solution for quality control in dECM production.