Red Blood Cell Phase‐image Segmentation with Deep Learning
Inkyu Moon · 2022
Deep learning is a promising technique that can offer results superior to those obtained by traditional methods. Convolutional neural networks are used for image classification with great success. Recurrent neural networks provide reasonably good performance for text classification and translation. Fully Convolutional Neural Networks (FCNs) are successfully applied to biomedical images such as cardiac segmentation in magnetic resonance imaging and liver or lesion segmentation in computed tomography. FCN has a translation-invariant feature due to the local connectivity properties of convolutional, pooling, Relu, and deconvolutional layers. The chapter introduces the Red Blood Cells (RBCs) phase-image segmentation procedure based on deep learning. Metrics of under-separating, over-separating, and encroachment errors were employed to quantitatively measure the RBC-separation abilities of these segmentation methods. Over-separating refers to the number of RBC divisions within a single, non-touching RBC.