1、 ANSI/CTA Standard Standard for Consumer EEG File Format (Attuned Container Format) ANSI/CTA-2060 November 2017 NOTICE Consumer Technology Association (CTA) Standards, Bulletins and other technical publications are designed to serve the public interest through eliminating misunderstandings between m
2、anufacturers and purchasers, facilitating interchangeability and improvement of products, and assisting the purchaser in selecting and obtaining with minimum delay the proper product for his particular need. Existence of such Standards, Bulletins and other technical publications shall not in any res
3、pect preclude any member or nonmember of the Consumer Technology Association from manufacturing or selling products not conforming to such Standards, Bulletins or other technical publications, nor shall the existence of such Standards, Bulletins and other technical publications preclude their volunt
4、ary use by those other than Consumer Technology Association members, whether the standard is to be used either domestically or internationally. Standards, Bulletins and other technical publications are adopted by the Consumer Technology Association in accordance with the American National Standards
5、Institute (ANSI) patent policy. By such action, the Consumer Technology Association does not assume any liability to any patent owner, nor does it assume any obligation whatever to parties adopting the Standard, Bulletin or other technical publication. This document does not purport to address all s
6、afety problems associated with its use or all applicable regulatory requirements. It is the responsibility of the user of this document to establish appropriate safety and health practices and to determine the applicability of regulatory limitations before its use. This document is copyrighted by th
7、e Consumer Technology Association (CTA) and the National Sleep Foundation (NSF) and may not be reproduced, in whole or part, without written permission. Federal copyright law prohibits unauthorized reproduction of this document by any means. Organizations may obtain permission to reproduce a limited
8、 number of copies by entering into a license agreement. Requests to reproduce text, data, charts, figures or other material should be made to the Consumer Technology Association (CTA) and the National Sleep Foundation (NSF). (Formulated under the cognizance of the CTA R6.4 Health Tel: +1 732 981 006
9、0; Internet: http:/dx.doi.org/10.1109/IEEESTD.2008.4610935, standards.ieee.org 2. The Unicode Consortium, Mountain View, CA; ISBN 978-1-936213-10-8); Internet: http:/www.unicode.org/versions/Unicode8.0.0 3. The World Wide Web Consortium (W3C); Internet: www.w3.org 4. International Organization for S
10、tandardization www.iso.org 2.2 Informative References The following references contain provisions that, through reference in this text, constitute informative provisions of this standard. At the time of publication, the edition indicated was valid. All standards are subject to revision, and parties
11、to agreements based on this standard are encouraged to investigate the possibility of applying the most recent edition of the standard indicated below. CTA-2060 2 2.2.1 Informative Reference List 1. The XDF (Extensible Data Format) Project 2.2.2 Informative Reference Acquisition 1. Swartz Center for
12、 Computational Neuroscience, 9500 Gilman Dr. #0559, La Jolla CA 92093-0559; Internet: http:/sccn.ucsd.edu/people/# and https:/ 2.3 Compliance Notation As used in this document “shall” and “must” denote mandatory provisions of the standard. “Should” denotes a provision that is recommended but not man
13、datory. “May” denotes a feature whose presence does not preclude compliance, and implementation of which is optional. “Optional” denotes items that may or may not be present in a compliant device. 2.4 Definitions Attuned Attuned Container Format 2.5 Symbols and Abbreviations ADC Attuned Container Fo
14、rmat BCI Brain-Computer Interface ECG Electrocardiography EEG Electroencephalography EMG Electromyography EOG Electroculography fMRI functional Magnetic Resonance Imaging HED Hierarchical Event Descriptor GSR Galvanic Skin Response MCN Modified Combinatorial Nomenclature for electrode placements in
15、the brain. MEG Magnetoencephalography MoCap Motion Capture nNIRS near-Infrared Spectrography XDF Extensible Data Format XML Extended Markup Language 3 Extended Data Format (Attuned Container Format) for EEG and multi-modal data storage 3.1 Introduction Attuned (Attuned Container Format) is an extens
16、ible storage format for time series data and associated metadata. Attuned was created to address the need for a true multi-modal time-series file format that can hold simultaneously multiple diverse time-series data streams. CTA-2060 3 A key requirement for Attuned was to be able to store essentiall
17、y any kind of time series data streams with few or no limitations on the number of channels, sampling rates, value formats, and amount or type of associated meta-data. A core use case for the format, is to store data from multiple sensors, which measure diverse physiological data (e.g., EEG and othe
18、r brain signals, heart rate, motion capture, eye tracking, video, audio, and human interface device inputs), but also any other kind of states that relates to such measurements (e.g., simultaneous application state and events). Attuned was designed with the additional requirements in mind that the f
19、ile contents shall be streamable (e.g., at the time of recording or playback), the contained streams can be time-synchronized, the file format is efficient in terms of storage space, and that the specification should be simple and compact. Based on these criteria, Attuned was designed as a chunk-bas
20、ed format. In a chunk-based format the entire file is a sequence of chunks, where multiple streams are stored in an interleaved, and therefore streamable fashion. This also enables forward compatibility, where implementations can skip over unknown chunk types (e.g., introduced in a future version of
21、 the standard) to allow for space-efficient storage of high-volume multi-channel time-series data. Attuned defines a chunk type that can store such data in a simple generic binary array format. These chunks also allow for storage of variable-length binary data, such as video frames, and therefore ca
22、n hold virtually any kind of streamable data in a storage-efficient manner. To satisfy the requirement for arbitrary extensible meta-data for each data stream, Attuned provides a flexible text-based chunk type that can hold any kind of structured metadata using XML as the encoding. While it is possi
23、ble to store arbitrary content in these chunks, to enable a maximum degree of compatibility Attuned recognizes several predefined types of stream content (“content-types“), such as EEG, Video, Audio, MoCap, Markers, etc., and for each content-type, a set of predefined metadata fields, such as channe
24、l labels, color space, units, etc., that are found in the appendix of this document. These fields were chosen based on the meta-data present in domain-specific “gold standard“ file formats (e.g., traditional video, EEG, or audio file formats). This appendix is expected to be amended in future revisi
25、ons of the standard to add support for additional content-types or to augment the set of standardized meta-data fields. It shall be emphasized however, that implementers (vendors) are free to add their own custom metadata fields or content-types to satisfy their storage needs. It is encouraged that
26、such extensions to be provided to a public Attuned extension registry. Unless indicated otherwise, registered extensions will automatically be considered for integration into subsequent Attuned revisions. Since fields are easy to add in future extensions, but hard to remove, the initial set of meta-
27、data fields tends to be conservative and covers primarily the essential and widely agreed-upon fields. To address the requirement for time synchronization in a fully general way, each sample has a timestamp, which is stored with an efficient compression scheme to minimize storage overhead. These tim
28、e stamps may even be read off of different clocks which are allowed to drift relative to each other. For such cases, the format may contain periodic clock offset measurements for each stream held in special type of chunk, which determine how these different clocks relate to each other at any given t
29、ime and allow for synchronization. While all time stamps could as well be stored in an already synchronized form, Attuned gives users the option to store the ground truth measurements, in order to not alter or falsify the raw data. An Attuned file can contain any number of (multi-channel) streams, e
30、ach of which comprises an XML-based header that specifies both format and meta-data of the stream, as well as a sequence of multi-channel data samples in a compact binary representation, which may also be empty if no sample data is present. CTA-2060 4 An implementation claiming Attuned 1.0 complianc
31、e must support the features described in Section 4, and may further support any number of registered or unregistered vendor extensions. This specification was evolved from XDF 1.0 specification. 3.2 Notation This document indicates named entities in the Attuned specification using Courier New font.
32、Non-normative sections that are not part of the specification, such as hints for implementers, are marked with an arrow character (). The binary layout of compound data in the file (e.g., chunks) is described in a three-row table. compound data (“compound items”) are items that consist of multiple “
33、sub-items”. The table defines the name of each subsequent sub-item in the first row, description (e.g., exhaustive lists or examples) in the second row, and the size of the sub-item in bytes in the third row. The following example defines a compound item with three sub-items, which comprise a 4-byte
34、 sub-item named Length, followed by a 2-byte sub-item named Length that encodes the length of the following sub-item, followed by a variable-length sub-item named Content that has arbitrary content. Name: Identifier Length Content Description: Identification number Content length in bytes Arbitrary
35、data Size (bytes): 2 4 Variable 3.3 Basic Concepts A data stream represented in Attuned may have a regular or an irregular sampling rate and each of its multi-channel samples can carry a time stamp in addition to its data values. The format of values in a data stream can be integers with 8, 16, 32,
36、and 64 bits in twos complement format (int8, int16, int32, int64), IEEE 754 32-bit and 64-bit floating point numbers (float32 and double64), and character sequences (string). All data values of a stream must be of the same format but different streams can have different data types. The time stamp fo
37、rmat is double precision in seconds). The Attuned format is (aside from an initial four-byte magic code) a sequence of chunks. The chunks listed below are detailed further within the document. The chunks that form a single stream are: one StreamHeader chunk (containing XML), zero or more Samples chu
38、nks (containing one or more samples of the stream), zero or more ClockOffset chunks (containing an offset for the time stamps of the stream used for synchronization), and zero or one StreamFooter chunk (which indicates that the stream was closed orderly). Furthermore, an Attuned file contains chunks
39、 not associated with a particular stream, namely a FileHeader chunk and optionally Boundary chunks that contain a special 16-byte sequence that help finding chunk boundaries when seeking into the file. CTA-2060 5 The first chunk in an Attuned file must be the FileHeader chunk, followed by meta-data
40、chunks (such as the StreamHeader chunks) in no particular order. After that, the Samples chunks follow, approximately sorted by time (that is, the time is approximately monotonically increasing), optionally interleaved with Boundary and/or ClockOffset chunks. At the end of the file, the StreamFooter
41、 chunks follow in no particular order. All binary fields (such as length fields) are stored as little-endian values. All numeric data types are also little-endian. All string types are assumed to be UTF8-encoded unless specified otherwise. Attuned uses XML-formatted strings as part of its specificat
42、ion. Generally, XML as used in the Attuned specification includes no attributes in the interest of simplicity. Attuned uses the file extension .adc. 4 Basic Structure of an Attuned file An Attuned file begins with the 4-byte ASCII-encoded magic code “adc:” and is followed by zero or more chunks. Nam
43、e: MagicCode Chunk Chunk Description: 4-byte character sequence adc Data chunk (see Section 5.1) Data chunk. Size (bytes): 4 Variable Variable 4.1 Chunk Chunks represent the only other content of an Attuned file besides the magic code. The length of each chunk is encoded as a variable-length integer
44、, the size of which is indicated by the first byte (which must be either 1, 4, or 8), followed by either a byte, a 4-byte unsigned integer, or an 8 byte unsigned integer. This is to support a uniform chunk format that can be used to hold both very short chunks and very long chunks without a large ov
45、erhead. The length refers to the length of the chunks remainder that follows the Length item. Name: NumLengthBytes Length Tag Content Description: Numbers 1, 4, or 8 Chunk length in bytes Tag number Arbitrary Size (bytes): 1 As coded in NumLengthBytes 2 Variable Notes for implementers: Tools reading
46、 Attuned files should skip chunk tags that they do not recognize or that they do not need to process. This is trivially enabled by the Length item, which makes it easy to skip chunks without inspecting their content. Also, tools CTA-2060 6 writing Attuned files should use the shortest encoding that
47、can hold the chunk data when writing. The chunk tag defines the type of the chunk. In the following sections, only the Content item for each of the respective chunk types is discussed. Currently used tags are 1-6, defined in the following: 1. FileHeader 2. StreamHeader 3. Samples 4. ClockOffset 5. B
48、oundary 6. StreamFooter Additional chunk types may be specified in future revisions of the Attuned standard and in vendor or general-interest extensions. Note for proposers of extensions: Any information that can be represented in any of the existing XML-formatted content areas is preferably stored
49、in these areas instead of in a special chunk type, since Attuned has a scalable process for defining such meta-data. Before a new tag is defined, the Attuned Container Format should be checked to avoid name collisions. Any newly added chunk must retain the NumLengthBytes, Length, and Tag elements so that programs can read through the chunks without having to parse the content structure. 4.2 FileHeader Chunk Name: HeaderInfo Description: Any Valid XML Size (bytes): As determined by chunk length This chunk must be at the beginning of the file, right af
