2017/07/19 by Aleksander Klibisz, Derek Rose, Klibisz, Aleksander +7 · 1 citation
Biochemistry, Genetics and Molecular Biology · Engineering · Materials Science · #Cell Image Analysis Techniques #Computer Vision and Pattern Recognition (cs.CV) #FOS: Biological sciences #FOS: Computer and information sciences #Image Processing Techniques and Applications #Machine Learning in Materials Science #Neurons and Cognition (q-bio.NC)
paper · pdf · doi:10.48550/arxiv.1707.06314
openalex publication_date 2017/07/19 · openalex created_date 2022/10/02 · openalex updated_date 2026/07/28
Calcium imaging is a technique for observing neuron activity as a series of\nimages showing indicator fluorescence over time. Manually segmenting neurons is\ntime-consuming, leading to research on automated calcium imaging segmentation\n(ACIS). We evaluated several deep learning models for ACIS on the Neurofinder\ncompetition datasets and report our best model: U-Net2DS, a fully convolutional\nnetwork that operates on 2D mean summary images. U-Net2DS requires minimal\ndomain-specific pre/post-processing and parameter adjustment, and predictions\nare made on full 512\×512 images at \≈9K images per minute. It\nranks third in the Neurofinder competition (F1=0.569) and is the best model\nto exclusively use deep learning. We also demonstrate useful segmentations on\ndata from outside the competition. The model's simplicity, speed, and quality\nresults make it a practical choice for ACIS and a strong baseline for more\ncomplex models in the future.\n