Technological improvements in hardware and computing power allow us to gather a larger volume of data. For example, larger amounts of measurements can be captured by a nuclear magnetic resonance spectrometer, scanning transmission electron microscope, or other scientific instrument, providing higher resolution for features, such as changes in the subject being measured, that might be represented in the data. Finding and determining the nature of a feature can involve manual review of thousands or more individual images, resulting in a time-consuming and sometimes futile process.

Pacific Northwest National Laboratory’s feature identification system can reduce or eliminate the need for manual review of a data set to identify or classify a feature, such as a change in measurements taken by a scientific instrument. Using metadata generated by a task-specific processing component, the patented software can detect a feature or event and generate a display or notification. For example, an event can be identified when metadata for one or more elements of the data set exceed one or more threshold values. The system facilitates and automates review processes, improving accuracy and precision and identifying features that would be difficult or impossible to classify manually using the raw data set.

The technology works with scientific instruments or other apparatus that generate information. For example, a camera can be used to record visual information regarding a variety of subjects. When used as a scientific instrument, the camera records visual information associated with the systematic study of a particular physical or natural process. Each piece of visual information is accompanied by metadata—data describing the image—which can be searched faster than reviewing the image itself.

The feature identification system can leverage a variety of data sets, including a set of discrete images, audio data, waveforms, or textual data. For example, the data set may include images of waveforms associated with a chemical reaction or process. A change in the waveform images can indicate a change in the reaction or process.

The technology comprises computer software to capture metadata, configured to the type of instrument or apparatus, as well as algorithms to analyze metadata. The system can also receive measurements recorded by an instrument, convert the measurements into a format usable for task-specific processing, and generate metadata for additional analysis.

Applicability

The ability to constantly monitor multiple streams of incoming data and only call human attention or use more computationally complex analysis when needed will be of value to multiple fields, such as:

• Event detection in streaming video, such as that from in situ (scanning) transmission electron microscopy (S/TEM)
• Event detection in streaming non-video data
• Shot boundary detection