SDE is a unique tool, containing a wide range of functionality, which provides invaluable assistance to analysts of time series data from many different commercial and academic sectors. It helps with the efficient exploration and investigation of large datasets and can automatically detect important trends and events. Full details of all the SDE functionality can be found in the SDE manual and help file, but as these are very lengthy and detailed documents, the following set of tutorials has been developed to assist new users in getting started with SDE.
Each of the tasks outlined in the tutorials is based on a different sample dataset. The datasets used are taken from a variety of the different SDE application areas, varying from ECG and EEG readings to electrical signal processing. However, all are designed so that they can be understood and completed without specialist knowledge in the subject areas and the techniques that will be taught will be applicable to any SDE application.
Visualising the data
This tutorial introduces the SDE interface and teach you how to open and display data effectively using the different data views available within SDE. After selecting an appropriate view, zoom level and play speed, you will replay an experiment that measured the response of retinal tissue to the presence of blue light. The SDE display provides a flexible, interactive representation of the experimental results, clearly showing how the spiking activity changes across the cellular array as the experiment progresses.
Exploring, Segmenting and Displaying data
Learn how to efficiently navigate and investigate several hours of high resolution, physiological data recorded during a patient’s sleep. By zooming out by a factor of up to 2000, any causes of disturbed sleep can be easily identified. You will be shown how to zoom in on these areas so that they can be investigated in more detail and how to create visual representations of these sections for use in presentations and reports. This tutorial will also explain how SDE can create new, smaller data files from selected segments of the currently opened data.
Define a spike/event by selecting an example template and specifying how similar in scale and shape potential spikes must be to this template, if they are to be considered a match. Use SDE to very rapidly locate any occurrences of the defined spike/event, replacing the analogue time series data with binary event data, indicating all the spike positions.
This tutorial will show you how to search for a broad range of spiking behaviour across multiple data channels and how to perform spike sorting, by independently selecting and classifying different spike shapes. Note that although this tutorial classifies the search procedure as spike detection, the techniques demonstrated can be similarly applied to other forms of event detection, such as the identification of ramps or steps.
SEARCHING FOR PATTERNS
Introduces the user to the main search facility within SDE, explaining the types of search available and the scenarios in which they are useful. This tutorial will show you how to locate the normal heartbeats in an ECG recording containing considerable irregular behaviour, whilst discussing each of the search parameters and the effect they have on the search performed. You will then be shown how to assess the results and improve the search and, once you are happy with the search definition, how to use or store the results obtained.
The second stage of this tutorial explains how the pattern finder facility can be used to create a library of important patterns and locate these with the current data file.
Controlling the SDE Search
Explains how the control the SDE search, using the available parameters and search options, to help ensure that the search returns the patterns in which you are interested. This includes an example of when to use the Similarity and when to use the Distance search, as well as explaining how to set the matching threshold and amplitude controls appropriately.
This tutorial will explain how you can apply a wide range of filters to your data and dynamically adjust the parameters, with instant visual feedback of the filter’s effect, to obtain the desired results. You will be shown how to improve your search results by smoothing the data prior to searching and, using the example of the construction of a full wave rectifier, how to chain multiple basic filters together to create new, more powerful filtering options.
Using Task Planner to Search for Coincident Activity
Learn how to use the task planner, available as part of the SDE professional package, to create more powerful search tasks, which can identify multiple, different, coincident patterns, across multiple variables. You will be taken through the full process of defining a series of tasks, which identify occurrences of coincident spiking activity across multiple variables.
Firstly you will investigate the data and plan how to effectively define the tasks. In stage 2 you will create the tasks using the task planner interface and then, in the final stage, you will assess the performance of your task library and make any improvements required to obtain perfect search results. An example of a perfectly functioning task library is available for download. This can be loaded into SDE if you are in need of assistance.
In order to use the task planner, the features on which you will be searching need to be defined in your variable configuration file. Details of the configuration file can be found in chapter 7 of the User Manual, but for this tutorial a feature_def.cfg file is supplied below. If you have downloaded the SDE version from the website, this configuration file will already be in place. However, if you have a bespoke version, you may need to copy the feature_def.cfg file below to the following locations before starting the final tutorial.
Copy the feature_def.cfg file to this location:
Note that these locations are often classified as hidden directories.