OpenIGTLink/ProtocolV2
!! PLEASE NOTE that this is a draft for protocol version 2, and might be changed a lot!!
Introduction
Overview
The Open IGT Link protocol is a simple but extensible data format to transfer various types of data among software and devices used in image-guided therapy (IGT) scene e.g. surgical navigation software, tracking device, robotic device, imaging scanner. The protocol can handle image, tracking data, transform, device control and monitoring command, and other user-defined data types. It is designed to work in the application layer of the TCP/IP stack, but not limited to it; developers can use this protocol with UDP/IP or other network protocol models.
The objective of providing this open network protocol is to make system integration for IGT application easier. To achieve this objective, we have the following rules to design the protocol:
- Open. It is important to invite the IGT developers as many as possible to make the protocol more universal, compatible and reliable. It is also important to allow commercial use without any fee, to increase the number of commercial softwares and devices compatible with Open IGT Link, which contributes interoperativity in the IGT scene.
- Simple. In IGT environment, wide variety of architectures are used, from embedded systems to high-performance computers. It is difficult to support all of them by providing a single open-source library, thus the developers sometime have to implement the interface by themselves.
- Extensible. Type of data transfered in the IGT scene varies by clinical applications. The protocol must have mechanism for the developers to define their own data types.
- Reliable. The protocol should provides a mechanism to verify data integrity. This is extremely important for the saftware to get approved and commercialized.
Protocol extensions
Client - Server
A device (scanner, tracker) exposing functionality and executing commands is usually the Server. Once started will stand by listening for connection requests.
A Client (for ex. a PC) usually connects to multiple devices and coordinates the whole process.
Common rules for data fields
Floating point data
Floating point values in the data must comply with IEEE 754-1985 standard.
Character string data
Character string (array) should end with '\0' (NULL), unless the string length is same as the length of the data field. For example when the length of the data field is 10 bytes:
0 1 2 3 4 5 6 7 8 9
+---+---+---+---+---+---+---+---+---+---+
"MONKEY" | M | O | N | K | E | Y |\0 | | | |
+---+---+---+---+---+---+---+---+---+---+
0 1 2 3 4 5 6 7 8 9
+---+---+---+---+---+---+---+---+---+---+
"CHIMPANZEE" | C | H | I | M | P | A | N | Z | E | E |
+---+---+---+---+---+---+---+---+---+---+
Header Structure
Bytes 0 2 14 34 42 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | V | TYPE | DEVICE_NAME | TIME_STAMP | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 42 50 58 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+..... | BODY_SIZE | CRC64 | BODY +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+.....
Byte Order
Big endian should be used for all numerical values (version, body size, crc). Unused spaces are padded with 0 (binary).
Header Fields
| Data | Type | Description |
| V | Unsigned short (16bit) | Version number (1) |
| TYPE | char[12] | Type name of data |
| DEVICE_NAME | char[20] | Unique device name |
| TIME_STAMP | 64 bit unsigned int | Timestamp or 0 if unused |
| BODY_SIZE | 64 bit unsigned int | Size of body in bytes |
| CRC | 64 bit unsigned int | 64 bit CRC for body data |
Data types (Common across implementations)
Data type:
- Command (get position, get status, go to target, stop, start continuous push, stop push, ready)
- Data (image, position, status)
POSITION
Quaternion representation of position / orientation.
Parameters (24 byte total):
- Vector of three 32 bit (3x4 byte) floats: position X, Y, Z
- Vector of four 32 bit (3x4 byte) floats: orientation quaternion, normalized
The orientation is optional, if not used the values are (0,0,0,1)
Bytes (Body) 58 60 64 68 72 76 80 84 88 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |V|O| X | Y | Z | OX | OY | OZ | W | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Data | Type | Description |
| V | 8-bit unsigned int | Format version number |
| O | 8-bit unsigned int | Coordinate system (1: RAS, 2:LPS) |
| X, Y, Z | 32-bit floating point | 3D poisition |
| OX, OY, OZ, W | 32-bit floating point | Orientation (in quaternion expression) |
LTRANSFORM
This data type represents 4x4 linear transformation matrix. The message contains upper three rows of 4x4 transformation matrix, where element (3,3) is assumed to be 1.
Each element is a 4 byte (32 bit) float, 12x4 = 48 byte in total.
Bytes (Body) 58 60 64 68 72 76 80 84 88 92 96 100 104 108 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |V|O|m(0,0) |m(1,0) |m(2,0) |m(0,1) |m(1,1) |m(2,1) |m(0,2) |m(1,2) |m(2,2) |m(0,3) |m(1,3) |m(2,3) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Data | Type | Description |
| V | 8-bit unsigned int | Format version number |
| O | 8-bit unsigned int | Coordinate system (1: RAS, 2:LPS) |
| m(i,j) | 32-bit floating point | Elements of 4x4 linear transformation matrix |
IMAGE
For normal scalar and vector images:
Bytes 58 60 62 64 70 82 94 106 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |V|O|T|S|E|R|RI |RJ |RK | TX | TY | TZ | SX | SY | SZ | NX | NY | NZ | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 106 118 124 130 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+.... | PX | PY | PZ |DI |DJ |DK |DRI|DRJ|DRK| IMAGE_DATA +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+....
For color-map images
Bytes 58 60 62 64 70 82 94 106 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |V|O|T|S|E|R|RI |RJ |RK | TX | TY | TZ | SX | SY | SZ | NX | NY | NZ | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 106 118 124 130 150 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | PX | PY | PZ |DI |DJ |DK |DRI|DRJ|DRK| COLOR_TABLE_NAME | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 150 +-+-+-+-+-+-+-+-+.... | IMAGE_DATA +-+-+-+-+-+-+-+-+....
| Data | Type | Description |
| V | 8 bit unsigned int | Format version number |
| R | 8 bit unsigned int | Reserved |
| T | 8 bit unsigned int | Image type (1: Scalar, 3: Vector, 17: Color-mapped Scalar, 19: RGB) |
| S | 8 bit unsigned int | Scalar type (2:int8 3:uint8 4:int16 5:uint16 6:int32 7:uint32 10:float32 11:float64) |
| E | 8 bit unsigned int | Endian for image data (1:BIG 2:LITTLE) (NOTE: values in image header is fixed to BIG endian) |
| O | 8 bit unsigned int | image coordinate (1:RAS 2:LPS) |
| RI, RJ, RK | 16 bit unsigned int | Number of pixels in each direction |
| PX, PY, PZ | 32 bit float | center position of the image |
| TX, TY, TZ | 32 bit float | Transverse vector (direction for 'i' index) / The length represents pixel size in 'i' direction |
| SX, SY, SZ | 32 bit float | Transverse vector (direction for 'j' index) / The length represents pixel size in 'j' direction |
| NX, NY, NZ | 32 bit float | Normal vector of image plane(direction for 'k' index) / The length represents pixel size in 'z' direction or slice thickness |
| DI, DJ, DK | 16 bit unsigned int | Starting index of subvolume |
| DRI, DRJ, DRK | 16 bit unsigned int | number of pixels of subvolume |
| COLOR_TABLE_NAME | char[20] (applicable if T == 17) | Name of color table. This field exists only when color-mapped scalar type is specified. |
| IMAGE_DATA | Binary image data () | Image data (endian is determined by "E" field) |
April 6, 2010: Exchanged "Reserved" and "Image coordinate" field to be compatible with version 1.
COLORTABLE
The COLORTABLE data type is used to transfer a table to convert scalar data to colored image. Color table data is transferred separately from image data; the receiver stores color tables with their names in the DEVICE_NAME fields of the generic headers. When receiver receives IMAGE data with color-mapped image option (T=17), it applies color table specified in the COLOR_TABLE_NAME field in the IMAGE message.
Bytes 58 60 62 +-+-+-+-+-+-+-+-+-+-+... |V|R|I|M| TABLE +-+-+-+-+-+-+-+-+-+-+...
| Data | Type | Description |
| V | 8-bit unsigned int | Format version number |
| R | 8-bit unsigned int | Reserved |
| I | 8-bit unsigned int | Index Type (3:uint8 5:uint16) |
| M | 8-bit unsigned int | Map Type (3:uint8 5:uint16 19: RGB color) |
| TABLE | Array of 8-bit unsigned int | Color index table |
The size of the TABLE field can be calculated by the following formula:
<size of TABLE> = <number of indices> * <pixel data size>
where
<number of indices> : 256 (if I=3) or 65536 (if I=5)
<pixel data size> : 1 (if M=3), 2 (if M=5) or 3 (if M=19)
GET_CAPABILITY
Query for device information and capabilities.
Parameters: none (BODY_SIZE=0)
Response: one CAPABILITY packet for each implemented device
CAPABILITY
List of available TYPEs.
Bytes (Body) 58 60 62 74 76 88 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+....-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |V|R|V|R| TYPE (12bytes) |V|R| TYPE (12bytes) | |V|R| TYPE (12bytes) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+....-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Data | Type | Description |
| V | 8-bit unsigned int | Format version number |
| R | 8-bit unsigned int | Reserved |
| TYPE | ASCII string (12 bytes) | Device Type name |
GET_STATUS
Parameters: none (BODY_SIZE=0)
Response: one STATUS packet for each implemented device
STATUS
Any STATUS that's not 1 (OK) should be logged by the receiver.
The uniform error reporting allows easy logging and can be the base of exception class in libraries.
Bytes (Body) 58 62 70 90 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+..... |V|R| C | Sub Code | Status name | Status message (optional) +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+.....
| Data | Type | Description |
| V | Unsigned short (8bit) | Version number |
| R | Unsigned short (8bit) | Reserved. |
| C | Unsigned short (16bit) | Status code groups: 1-Ok, 2-Generic Error, ... (see below) |
| Sub Code | 64 bit integer | Sub code for the error (ex. 0x200 - file not found) |
| Error name | char[20] | "Error", "OK", "Warning" - can be anything, don't relay on this |
| Status Message (optional) | char[ BodySize - 30 ] | Optional (English) description (ex. "File C:\test.ini not found") |
Status codes:
0 Invalid packet - 0 is not used
1 OK (Default status)
2 Unknown error
3 Panic mode (emergency)
4 Not found (file, configuration, device etc)
5 Access denied
6 Busy
7 Time out / Connection lost
8 Overflow / Can't be reached
9 Checksum error
10 Configuration error
11 Not enough resource (memory, storage etc)
12 Illegal/Unknown instruction (or feature not implemented / Unknown command received)
13 Device not ready (starting up)
14 Manual mode (device does not accept commands)
15 Device disabled
16 Device not present
17 Device version not known
18 Hardware failure
19 Exiting / shut down in progress
User-defined Data Types
Developers can define their own type of data. As long as the general header contains correct size information, OpenIGTLink-compliant software and hardware can at least skip the data and interpret following data correctly.
- Back to OpenIGTLink
