File format of Scanditronics/GEMS on-line detector raw data

Specifications

The data collected using the ABSS ("blood pumps" with on-line detectors) is saved in ASCII files named as *.bld or *blo.lis. The format is specified by the manufacturer of the devices, except for the header lines, starting with #; these lines are written by in-house programs. These lines must specify at least the date of study in a format shown in the example below, because the applied calibration coefficients are selected based on this date. Into the oldest data files the date has to be added manually.

The data columns are:

  1. Count collection start time as seconds from the beginning of the day
  2. Time (sec) from the start of this study
  3. Measurement interval, i.e. count collection time (sec)
  4. Coincidence counts from the first pair of detectors
  5. Single counts collected by the first detector of the first pair
  6. Single counts collected by the second detector of the first pair
  7. Coincidence counts from the second pair of detectors
  8. Single counts collected by the first detector of the second pair
  9. Single counts collected by the second detector of the second pair
  10. Aux counts (always zero in our measurements)

Example #1: GEMS on-line detector raw data file from a [15O]O2 study

# Protocol: /usr/PET/frq/frqflow5.prt
# Patient: ut193
# Isotope half-life: 2.05
#                 Start   Interv       1st detector pair       2nd detector pair     AUX
#                  time     time   coinc  singl1  singl2   coinc  singl1  singl2  counts
# 2002-06-25 12:59:04
ut193 2.050000 1.230000 1.400000 1.400000 1.400000 1.400000
  46834.0       0.0      1.0       5     877     783      15    1505    1864       0
  46835.0       1.0      1.0       3     719     511      13    1103    1254       0
  46836.0       2.0      1.0       4     650     503       8    1005    1166       0
  46837.0       3.0      1.0       3     632     468       8    1006    1094       0
  46838.0       4.0      1.0       2     639     511       6    1005    1063       0
  46839.0       5.0      1.0       4     589     461       9     987    1030       0
  46840.0       6.0      1.0       3     543     431       5    1012    1015       0
  46841.0       7.0      1.0       3     613     479       8     995    1126       0
  46842.0       8.0      1.0       2     621     441       8     928    1079       0
  46843.0       9.0      1.0       3     655     495       6     915    1105       0
  46844.0      10.0      1.0       9     658     504       8     973    1106       0
  46845.0      11.0      1.0      19     731     557      25    1023    1150       0
  46846.0      12.0      1.0      92     955     810      74    1215    1273       0
  46847.0      13.0      1.0     165    1229    1105     166    1544    1547       0
  46848.0      14.0      1.0     237    1484    1316     220    1647    1704       0
  46849.0      15.0      1.0     287    1750    1604     338    1988    2067       0
  46850.0      16.0      1.0     353    1978    1811     366    2128    2216       0
  46851.0      17.0      1.0     406    2151    1952     424    2279    2336       0
  46852.0      18.0      1.0     360    2183    2052     459    2351    2435       0
  46853.0      19.0      1.0     410    2269    2073     475    2424    2561       0
...

Example #2: GEMS on-line detector raw data file after control software was updated to an in-house program

# GEMS Automated Blood Measurement System
# Protocol: 180 1
# Patient: tf04042018_2
# Isotope half-life: 109.8
#                 Start   Interv       1st detector pair       2nd detector pair     AUX
#                  time     time   coinc  singl1  singl2   coinc  singl1  singl2  counts
# 2018-04-04 11:31:33
      41493.5       0.0      1.0     283     849     931     230     877     652       0
      41494.5       1.0      1.0     259     851     823     237     824     694       0
      41495.5       2.0      1.0     289     897     909     231     912     682       0
...

Example #3: Scanditronics on-line detector raw data file from a [15O]O2 study

The first column contains the time in seconds since the epoch in local time.

# Scanditronics Automated Blood Measurement System
# Protocol: 400 1
#                 Start   Interv       1st detector pair       2nd detector pair     AUX
#                  time     time   coinc  singl1  singl2   coinc  singl1  singl2  counts
# 2002-02-06 11:01:35
 1012986129.0       1.0      1.0       0     114     152       5     634     333       0
 1012986130.0       2.0      1.0       0     133     176       6     668     332       0
 1012986131.0       3.0      1.0       3     134     182       5     680     350       0
 1012986132.0       4.0      1.0       2     132     183       4     742     309       0
 1012986133.0       5.0      1.0       3     184     263      10     793     437       0
 1012986134.0       6.0      1.0       3     188     227      20     866     461       0
 1012986135.0       7.0      1.0      16     196     276      17     823     478       0
 1012986136.0       8.0      1.0      20     232     300      18     944     504       0
 1012986137.0       9.0      1.0      20     244     338      39    1083     579       0
 1012986138.0      10.0      1.0      41     345     446      58    1204     606       0
 1012986139.0      11.0      1.0      63     397     471      81    1322     773       0
 1012986140.0      12.0      1.0      68     484     582      96    1444     843       0
 1012986141.0      13.0      1.0     102     617     658     126    1645     934       0
 1012986142.0      14.0      1.0     131     665     816     172    1864    1097       0
 1012986143.0      15.0      1.0     152     873     958     219    2023    1274       0
 1012986144.0      16.0      1.0     187     978    1102     264    2385    1447       0
 1012986145.0      17.0      1.0     243    1162    1252     334    2680    1620       0
...

Calibration of data

Each measurement line is calibrated to units kBq/ml by following this procedure. Note that this does not contain correction for physical decay or dispersion. Normally, blo2kbq or absscal can be used for this purpose.

  1. Calculate the average of coincidence counts (columns #4 and #7)
  2. Divide the average by measurement duration in seconds (column #3)
  3. Read the two calibration coefficients from calibration coefficient file:
    • selected detector specific detector-to-gamma counter calibration coefficient: “pump1(ecat)” for *blo.lis files and “pump2(ge)” for *.bld files
    • read also the last column, gamma counter-to-pet coefficient, which is common to all on-line detectors
    • select calibration that is made just before or on the same day as the blood measurement
  4. Multiply the two calibration coefficients with each other
  5. Divide the calibration coefficient by the branching ratio of the radioactive label isotope
  6. Multiply the count-per-second value from step #2 by the combined calibration coefficient

Sample time for each measurement is calculated as the sum of measurement start time (column #2), and measurement duration (column #3) divided by 2, i.e. the middle point of each measurement. There are separate instructions for correcting data for the physical decay.

Problems with on-line detector raw data

"Dead" detector pair

In some cases, either of the pairs can not calculate the coincidences and print zeros the 4th or 7th column in raw data file. This results in blood TAC files that are about one-half of the normal curves. If you suspect or have found out that this has happened, contact Tuula Tolvanen (or try to use abssfch).

Interrupted and remade study

If the same study has been started again because of some failure in the first try, it may happen that on-line detector raw data consists of the start of the interrupted study, and directly after it, the data from the successful second try. In this case, you can read the raw data file in a text editor, remove there the data belonging to the first study, and then save the data again in ASCII file, and then continue the analysis as usual. However, you should make a backup copy of the original raw data file before editing it.

Scanditronics data collected before Oct 1998

The original VAX software was used to run the Scanditronics on-line sampler at ECAT 931 before Oct 4, 1998. Data files from that time do not contain any title lines and are in VAX format, which can not be processed by the recent software releases. If you need to re-analyze old data, please try to use previously calibrated blood data.

Scanner and blood sampler in different time

In early 2006, the time in PET-CT was not yet the same as in other equipments (computer clocks were not synchronized). The different time must be corrected to the online sampler data, before it is processed further.

If PET scan and blood sampler were started simultaneously, the start times in dynamic PET image and in blood sampler data file must be the same. This can be checked using egetstrt and absstime and, if necessary, also corrected using absstime.

Faulty sample times in Scanditronics data

Sample times (first columns) are wrong in Scanditronics data collected with bsampler 2.2.0. However, the measurement start time in file header is correct. blo2kbq (and absscal) uses correct sample start time from the header.

Blood sampler curve has a high initial activity

High activity values in the beginning of the blood curve will lead to error later in the time delay correction, and should be corrected either manually with a text editor or with absszero.

Not really problems, but things to be noted:

Scanditronics count collection start time

Count collection start time (column 1), has not been recorded as seconds from the beginning of the day, but as seconds from 1970-01-01 00:00:00 with certain bsampler releases. This does not affect the results.

Start time with GEMS on-line sampler

Data that has been collected with GEMS on-line sampler has a date and time in its header section. However, this is NOT the time at when the measurement actually started, but this is the time when the blood collection software began to wait for the actual scan start. If measurement start time is needed, it can be read from the first sample line, the first column.

In Scanditronics data files the start time, if it exists in the header, is the actual start time, and can be used as such.



Tags: , , , ,


Created at: 2006-10-13
Updated at: 2016-12-14
Written by: Vesa Oikonen, Tuula Tolvanen