Why does storage matter in bioimaging?
Microscopy experiments create digital images (pixel data) accompanied by accessory data about the image acquisition settings, acquisition date, etc. (metadata). Inside the computer, this information is stored in binary code as a series of 1s and 0s (see also Microscopy File Formats). To wrap all the information contained in a microscopy image recording together, the software uses a specific file format that defines how the 1s and 0s are arranged and organized to faithfully represent the recorded image and its metadata.
Modern microscopy produces files that contain an extremely large number of 1s and 0s, typically organized into arrays of data along multiple axes (x, y, z, time, channels). These are large, N-dimensional array-typed data, also known as tensors. Depending on the modality, the size of an imaging data file ranges from 10s to 100s of MB (e.g., a confocal image) to GB-scale images (e.g., whole-slide scanning) and up to TB-scale in some cases (e.g., volume electron microscopy). In comparison, photos taken with modern cameras typically range at ~20-30 MB even in the uncompressed raw format, which gets reduced to often below 10 MB in compressed JPEG images. This alone tells us: Microscopy images are much richer in information than „normal“ photos.
The large file sizes in bioimaging have two consequences for practical purposes:
- There must be sufficient storage hardware to contain all the 1s and 0s organized into files and keep this information faithfully contained over time.
- The storage hardware must enable computers to access this data with sufficient speed to process the content through computer programs – be it to visualize the data or to perform some analysis with the data. In other words, besides file formats, also storage hardware types matter significantly.
Back to: Storage
