Dumbbell, crescent and tadpole shape deformability of erythrocytes on intravascular photobiomodulation via micro-pillar array with deep learning visualizing

Abstract

An advanced detection system of erythrocytes integrated with the micro-pillar array in the microfluidic chip (MPMC) and deep learning visualizing system (DLVS) has been developed to measure the deformability of stroke patients’ erythrocytes before and after the intravascular photobiomodulation (iPBM). The pass-through rate of healthy individuals and stroke patients’ erythrocytes can be controlled via tuning the spacing of micro-pillars and calculated through dividing the number of erythrocytes at the outlet by the number at the inlet. DLVS evaluates erythrocytes’ health condition with a convolutional neural network and software to replace manual counting, which helps to improve the overall detection efficiency. When the spacing of micro-pillars is down to 3.21 μm, the stroke patients’ erythrocytes after iPBM transform squeezing into the micro-pillar from dumbbell to crescent and tadpole-like shape, expecting that their stiffness is modulated to become extremely low compared to those patients without iPBM. The pass-through rate change of stroke patients before and after iPBM is about 15–20 %. The shrinkage rate of RBC areas is from 20 to 44 %, and the deformability change evaluated by circularity is from 47 % to 62 % depending on specific shapes, respectively. High crescent-shape deformability of erythrocytes treated by iPBM causes the higher pass-through rate to facilitate erythrocytes to squeeze into the micro-pillar array. This advanced detection system is a breakthrough development that allow stroke patients to realize their blood health condition and doctors to diagnose patients’ future clinical treatment precisely.