Abstract
Cell image analysis is crucial in Alzheimer's research to detect the presenceof A$\beta$ protein inhibiting cell function. Deep learning speeds up theprocess by making only low-level data sufficient for fruitful inspection. Wefirst found Unet is most suitable in augmented microscopy by comparingperformance in multi-class semantics segmentation. We develop the augmentedmicroscopy method to capture nuclei in a brightfield image and the transformerusing Unet model to convert an input image into a sequence of topologicalinformation. The performance regarding Intersection-over-Union is consistentconcerning the choice of image preprocessing and ground-truth generation.Training model with data of a specific cell type demonstrates transfer learningapplies to some extent. The topological transformer aims to extract persistence silhouettes orlandscape signatures containing geometric information of a given image ofcells. This feature extraction facilitates studying an image as a collection ofone-dimensional data, substantially reducing computational costs. Using thetransformer, we attempt grouping cell images by their cell type relying solelyon topological features. Performances of the transformers followed by SVM,XGBoost, LGBM, and simple convolutional neural network classifiers are inferiorto the conventional image classification. However, since this researchinitiates a new perspective in biomedical research by combining deep learningand topology for image analysis, we speculate follow-up investigation willreinforce our genuine regime.