AI-based autofocusing of red blood cells in digital in-line holographic microscopy

Autofocusing methods have been commonly utilized to measure depth positions of red blood cells (RBCs) from their holographic images captured by a digital in-line holographic microscopy (DIHM) system. However, conventional autofocusing methods have technical limitations in automatically and accurately determining the center of mass of biconcave RBCs positioned at different orientations. In this study, an artificial intelligence (AI)-based autofocusing method is proposed for precise measurement of depth positions of RBCs. Holographic images of RBCs captured at different depth positions and their corresponding focus value labels are used to train a convolutional neural network (CNN) model. The trained CNN model accurately evaluates focus values of RBC holograms, enabling the determination of depth positions of RBCs. Comparative analysis with conventional methods demonstrates superior autofocusing performance of the proposed AI-based autofocusing technique. The proposed technique would be utilized to analyze 3D dynamic behaviors of RBCs in complex 3D flows under various microfluidic conditions.