Deep-Learning Based Estimation of Dielectrophoretic Force

Abstract

The ability to accurately quantify dielectrophoretic (DEP) force is critical in the development of high-efficiency microfluidic systems. This is the first reported work that combines a textile electrode-based DEP sensing system with deep learning in order to estimate the DEP forces invoked on microparticles. We demonstrate how our deep learning model can process micrographs of pearl chains of polystyrene (PS) microbeads to estimate the DEP forces experienced. Numerous images obtained from our experiments at varying input voltages were preprocessed and used to train three deep convolutional neural networks, namely AlexNet, MobileNetV2, and VGG19. The performances of all the models was tested for their validation accuracies. Models were also tested with adversarial images to evaluate performance in terms of classification accuracy and resilience as a result of noise, image blur, and contrast changes. The results indicated that our method is robust under unfavorable real-world settings, demonstrating that it can be used for the direct estimation of dielectrophoretic force in point-of-care settings.

Comparing machine learning and deep learning regression frameworks for accurate prediction of dielectrophoretic force

“…In 16 negative DEP of the PS beads was observed when a voltage of 3.8 V pp of 480 kHz frequency was applied, forming 6–7 beads-long pearl chain. Similar studies on yeast cells have been reported when voltage (3.7 V pp ) at a field frequency of 580 kHz exhibited positive DEP, the number of particles aggregated were found to be related to the applied voltage 16 , 17 .…”

“…The interaction of a non-uniform electric field with a dipole is known as dielectrophoretic force .To improve the approximation of the exerted on the particles in terms of the voltage applied, a more tangible and straightforward model is required to advance DEP aided sensing systems. The when the particle is significantly smaller than the non-uniformities in the electric field is given in 17 as: where is the radius of spherical microparticles in a medium, under an alternating current (AC) field ; depends on the product of the localized field with its gradient ( ) and the frequency-dependent complex dielectric contrast of the particle versus the medium, as given by real-part of the Clausius–Mossoti factor ; …”

“…The maximum attractive and repulsive forces in a chain grow significantly with the number of particles in the chain, but when the number of particles is large enough, they reach saturation. This was analytically calculated and given as 17 : where the relative permittivity of the medium, E is the electric field on the microparticle surface, is the normal component of E , and n is the unit normal vector on the surface. However, theoretical estimate of from micrographs is problematic due to discrepancies in the thread structure.…”

“…In our previous work 12 , 17 , we have reported the use of textile electrode-based DEP device in conjunction with deep learning to predict the induced on PS microbeads where we cast the estimation task as a classification problem. Convolutional Neural Networks (CNNs) classifiers were used to estimate the applied DEP voltage from the pearl chain alignment of the microbeads.…”

“…This was done to tackle the limitations in the existing estimation techniques such as the equivalent dipole (EDM) 39 , 40 , Maxwell stress tensor (MST) 41 – 44 , iterative dipole moment (IDM) 9 , 40 , 45 , 46 or velocity tracking method. We had used a classification-oriented CNN method in training the CNN models, which though gave excellent training results, performed poorly on testing with adversarial samples 12 , 17 . Also, classification paradigm may be inadequate for accurate computation as the spacing between distinct voltages is crucial.…”