Common Tracing Output Format

Contents 1 Introduction 2 SWC Format 3 Assumptions 4 Tracing Output in Compliance with SWC 5 Example 6 Sample XML 7 References

Introduction

In this page we describe a common output format for tracing algorithms in FARSIGHT. This applies to images with tubular-shaped structures such as micro blood vessels, neuronal arbor (with or without soma(s)) or spiny neuronal dendrites. In an effort to promote standardization, this format complies with SWC ^[1]. For the output of any tracing program to be readable by FARSIGHT, compliance with this specification is necessary. The final output has to be in XML.

SWC Format

This format is described in detail in Ascioli et al.'s work.^[1] The image topology has to be preserved regardless of the tracing method. Moreover, because each tracing algorithm may produce different ordering of center-points, identifying soma and/or branch points remains the only anchors that preserve the global topology. Therefore, any available soma or branch points are assumed to be segmented as part of the algorithm output. The topology is maintained by using directional connections between the anchor points and the traced structures.

Every SWC line (representing a dendrite, blood vessel, or a part thereof) is a potential parent and has

1. a unique ID (1st column or field)
2. a type descriptor (2nd column T: 1=Soma, 3=dendrite)
3. the x,y,z coordinates of the starting point
4. a Connectivity index (to which this segment is connected in a direction leading to the tree top
   • The tree top is either the soma (if available) or the starting line.
   • The tree top has ID=1 and connectivity index= -1
   • A tracing segment may connect back to the top (ID=1) or to another segment via its ID.

Assumptions

The tracing algorithm is assumed to segment / detect the following structures (whenever applicable) and include them in the output:

• soma and its branches
• dendritic branch points accurate to the tracing segment resolution
• dendritic spines

Tracing Output in Compliance with SWC

1. The soma (if present) is considered the highest in the structural hierarchy. It is listed with type=1, and connectivity=-1, its centroid coordinates and radius.
2. Next comes the traceline (also blood vessel or dendrite) listed with its ID, connectivity to its parent ID (the top parent=-1 if no soma is available), and type=3. The coordinates are given by its children listed right after it.
3. The traceline children are the tracebits (trace points or center-line points). They are listed with their parent type and ID, x,y,z coordinates. The tracing algorithms in FARSIGHT may list more tracebit details than the minimal SWC requirements such as radii, orientation, foreground intensity ...
4. Dendritic spines: FARSIGHT's spine segmentation algorithms associate a tracebit with every spine and provide the spine center-lines as part of the output. We decided to utilize this conveniently and treat spines as regular branches from tracelines whereas the branching point is the tracebit associated with each spine.

Hence, a spine is represented as a traceline of type=5 and whose tracebits are its center points.

(Example from Fig1 below: ID=2 is a dendrite segment connecting to soma; ID=8 is a dendrite segment connecting to ID=4- a dendrite segment.)

Parent types are Soma (S) and Dendrite (D), (may add more if needed). Tags are soma, traceline, tracebit, spine ID and object tag identify the XML line (object) uniquely. The IDs are unique only within their tree/type level. If there is one or more soma, each has its line, type, and ID: <Soma ID=1 ParentID=-1 X=-1.5 Y=-8.2 Z=1.3 Width1=3.8> A dendrite (tag= TraceLine) is a collection of segments (tag=TraceBit) A dendrite may branch off of a soma or another dendrite. This is identified by its ParentType=S or D, and ParentID=number. If the ParentType=D, then it must have branched from a segment at BranchTraceBit=number. Each dendrite segment (tag=TraceBit) has its unique ID within the TraceLine.

Example

Sample XML

References

1. ↑ ^{1.0 1.1} Giorgio A. Ascoli, Jeffrey L. Krichmar, Slawomir J. Nasuto, Stephen L. Senft. "Generation, Description and Storage of Dendritic Morphology Data", Philosophical Transactions: Biological Sciences, Vol. 356, No. 1412, Neuroscience Databases: Tools for Exploring Brain Structure-Function Relationships (Aug. 29, 2001), pp. 1131-1145