Features extracted from BAFET


After traces are marked in BAFET, the last step is outputting features based on topology and intensity. Three tables are used to record detailed features from each case.

feature: summary of each cases with 42 features in topology and intensity

vessel: detailed 17 features for each vessel in each case

vessel_group: vessels in each case categorized in 5 groups with similar vessels in same group, each group with 12 features


In features table:

Patient name: the name of the patient acquired from DICOM information stored in image set

Patient age: the age of the patient acquired from DICOM information stored in image set 

Patient gender: the name of the patient acquired from DICOM information stored in image set

Circulation Type: type for identifying circulation this case describes. 1 for anterior, 2 for posterior

Artery section area: the average section area of all artery seeds (bottom node in an artery)

Artery seeds: number of artery seeds in a case

Total Length: total length of all vessels (in the unit of pixel)

#Branches: number of vessel branches (a trace whose first node is in another trace)

#Bifurcations: number of vessel bifurcations (node whose number of neighbor node is at least three)

#Tips: number of end point (node whose neighbor node number is only one)

Average Diameter: average diameter of all vessels

Average Section Area: average section area of all vessels

Volume: total volume vessels take in image set

Surface: total area of vessel surface 

Average Branch Order: average order of vessel in a case

Average Bifurcation Angle: average value of minimum angle of branches in bifurcations

1st Intensity: average raw intensity value in vessels whose group (see vessel group table below) is 1

2nd Intensity: average raw intensity value in vessels whose group is 2

3rd Intensity: average raw intensity value in vessels whose group is 3

4+ Intensity: average raw intensity value in vessels whose group is 4

1st Length: total length of vessels whose group is 1

2nd Length: total length of vessels whose group is 2

3rd Length: total length of vessels whose group is 3

4+ Length: total length of vessels whose group is 4

1st Volume: total volume of vessels whose group is 1

2nd Volume: total volume of vessels whose group is 2

3rd Volume: total volume of vessels whose group is 3

4+ Volume: average length of vessels whose group is 4

MinSnakeIntensity: Minimum intensity of snake area

MaxSnakeIntensity: Maximum intensity of snake area

MedianSnakeIntensity: Median intensity (histogram bin (bin size of 1) location just over 50%)  of snake area

MeanSnakeIntensity: Mean/Average intensity of snake area

SigmaSnakeIntensity: Standard deviation of instensity in snake area

SumSnakeIntensity: Total intensity of snake area

SnakePixels: Number of pixels of snake area

MeanImageIntensity: Mean/Average image intensity of whole image

SigmaImageIntensity: Standard deviation of intensity of whole image

MinImageIntensity: Minimum intensity of whole image

MaxImageIntensity: Maximum intensity of whole image

MedianImageIntensity: Median intensity (histogram bin (bin size of 1) location just over 50%)  of whole area

5%ImageIntensity: histogram bin (bin size of 1) location just over 5%  of whole area

95%ImageIntensity: histogram bin (bin size of 1) location just over 95%  of whole area


In vessel table:

Patient name: the name of the patient acquired from DICOM information stored in image set. use as key to join tables in database.

Vessel Id: id of vessel starting from 0

StartPointLandmarkType: Type of landmark type (see 'Landmark type' below) of start point of vessel

EndPointLandmarkType: Type of landmark type of end point of vessel

StartPointPosition: x y z value of start point

EndPointPosition: x y z value of end point

VesselType: type of vessel (see 'Vessel type' below)

NumPts: number of points in describing this vessel

TotalLength: length in pixel of this vessel

A_Diameter: average (based on points) diameter of this vessel

A_SectionArea: average (based on points) area in cross section area of this vessel

Volume: total volume (in pixel^3) of this vessel

A_Intensity: average (based on points) intensity of value in this vessel

Surface: total surface in this vessel

Tortuosity: length of vessel divided by direct length from start point to end point

Bifurcation Angle: Minimum angle of all the angles in birfurcation area

Order: Minimum number of bifurcations this vessel meets when connecting to ICA root


In vessel group table:

Patient name: the name of the patient acquired from DICOM information stored in image set. use as key to join tables in database.

GroupType: id of vessel group starting from 0 to 5 (see 'Group type' below). 

NumPts: number of points in describing all vessels in this group

NumBranches: number of vessels in this group

TotalLength: total length of all vessels in this group

Volume: total volume of all vessels in this group

A_Diameter: average (vessel length weighted) diameter of all vessels in this group

A_SectionArea: average (vessel length weighted) cross-section area of all vessels in this group

A_Intensity: average (vessel length weighted)  intensity of all vessels in this group

Surface: total surface of all vessels in this group

Tortuosity: Average (vessel length weighted) tortuosity

Bifurcation: Average (unweighted) bifurcation angle


Landmark type:

"/" means connected point of two or more vessels

LandmarkName[0]="Undefined";  

LandmarkName[1]="ICARoot_L"; 

LandmarkName[2]="ICARoot_R"; 

LandmarkName[3]="ICA/MCA/ACA_L"; 

LandmarkName[4]="ICA/MCA/ACA_R";

LandmarkName[5]="A1/A2_L";

LandmarkName[6]="A1/A2_R";

LandmarkName[7]="M1/M2_L";

LandmarkName[8]="M1/M2_R";

LandmarkName[9]="OA/ICA_L";

LandmarkName[10]="OA/ICA_R";

LandmarkName[11]="OA_L";

LandmarkName[12]="OA_R";

LandmarkName[13]="M2/M3_L";

LandmarkName[14]="M2/M3_R";

LandmarkName[15]="VA_L";

LandmarkName[16]="VA_R";

LandmarkName[17]="BA/VA";

LandmarkName[18]="PCA/BA";

LandmarkName[19]="P1/P2/Pcomm_L";

LandmarkName[20]="P1/P2/Pcomm_R";

LandmarkName[21]="Pcomm/ICA_L";

LandmarkName[22]="Pcomm/ICA_R";


Vessel type: 

VesselName[1]="ICA_L";

VesselName[2]="ICA_R";

VesselName[3]="M1_L";

VesselName[4]="M1_R";

VesselName[5]="M2_L";

VesselName[6]="M2_R";

VesselName[7]="A1_L";

VesselName[8]="A1_R";

VesselName[9]="A2_L";

VesselName[10]="A2_R";

VesselName[11]="AComm";

VesselName[12]="M3_L";

VesselName[13]="M3_R";

VesselName[14]="VA_L";

VesselName[15]="VA_R";

VesselName[16]="BA";

VesselName[17]="P1_L";

VesselName[18]="P1_R";

VesselName[19]="P2_L";

VesselName[20]="P2_R";

VesselName[21]="PComm_L";

VesselName[22]="PComm_R";

VesselName[23]="OA_L";

VesselName[24]="OA_R";


Group type:

1 for ICA  

2 for M1 and A1

3 for M2, A2, Acomm, Pcomm, OA

4 for M3

0 for any vessels not labelled, Excluding posterior vessels such as P1, P2, BA, VA, which are to be grouped in other methods in the future


These features described comprehensive information in a image set. I believe they are able to reflect vessel richness and other characters for the help of brain health analysis, which is proved in Feature validation based on appearance tags

Last Article Next article

Comment 评论



Share 分享

New Users 最新加入

  • hokurikustr

  • refrain

New comments 最新评论

test123: aasdas Details Apr 13 16:39
admin: Thanks! Details Apr 09 11:46
admin: Google map api Details Apr 09 11:46
lqj12: cooooooooool Details Apr 08 21:34
Yunhan Huang: 这个功能是如何实现的? Details Apr 08 13:23