{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/deep-learning-approach-to-fourier","title":"Deep learning approach to Fourier ptychographic microscopy","arxiv_id":"1805.00334","date":"2018-04-27","proceeding":null,"authors":["Thanh Nguyen","Yujia Xue","Yunzhe Li","Lei Tian","George Nehmetallah"],"abstract":"Convolutional neural networks (CNNs) have gained tremendous success in\nsolving complex inverse problems. The aim of this work is to develop a novel\nCNN framework to reconstruct video sequence of dynamic live cells captured\nusing a computational microscopy technique, Fourier ptychographic microscopy\n(FPM). The unique feature of the FPM is its capability to reconstruct images\nwith both wide field-of-view (FOV) and high resolution, i.e. a large\nspace-bandwidth-product (SBP), by taking a series of low resolution intensity\nimages. For live cell imaging, a single FPM frame contains thousands of cell\nsamples with different morphological features. Our idea is to fully exploit the\nstatistical information provided by this large spatial ensemble so as to make\npredictions in a sequential measurement, without using any additional temporal\ndataset. Specifically, we show that it is possible to reconstruct high-SBP\ndynamic cell videos by a CNN trained only on the first FPM dataset captured at\nthe beginning of a time-series experiment. Our CNN approach reconstructs a\n12800X10800 pixels phase image using only ~25 seconds, a 50X speedup compared\nto the model-based FPM algorithm. In addition, the CNN further reduces the\nrequired number of images in each time frame by ~6X. Overall, this\nsignificantly improves the imaging throughput by reducing both the acquisition\nand computational times. The proposed CNN is based on the conditional\ngenerative adversarial network (cGAN) framework. Additionally, we also exploit\ntransfer learning so that our pre-trained CNN can be further optimized to image\nother cell types. Our technique demonstrates a promising deep learning approach\nto continuously monitor large live-cell populations over an extended time and\ngather useful spatial and temporal information with sub-cellular resolution.","url_abs":"http://arxiv.org/abs/1805.00334v3","url_pdf":"http://arxiv.org/pdf/1805.00334v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"deep-learning-approach-to-fourier","repo_url":"https://github.com/32nguyen/DeepLearningFourierPtychographicMircoscopy","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":null,"task_name":"Generative Adversarial Network"},{"task_slug":"time-series","task_name":"Time Series Analysis"},{"task_slug":"transfer-learning","task_name":"Transfer Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}