BMMR2: Breast Multiparametric MRI for prediction of NAC Response Forum

Go back to competition Back to thread list Post in this thread

> Data Format

Hello,I am a novice to processing MR images. I am confused about the naming of the image. I want to know which images are the original DWI data, T2-weighted images, DCE images with derived enhancement maps. If possible, please explain the naming rules. Thank you very much for your help.

1.000000-ISPY2 3 Plane Scout-75696
3.000000-ISPY2 WATERT2fseidealarcBP-19031
5.000000-ACRIN-6698 4bval100260048008T22nexeDWI-39152
6.000000-ISPY2 multiPhase38425680-100sec.Dyn.-72379
399.000000-ISPY2 T2fseidealarcBP-27539
3992.000000-ISPY2 WaterT2fseidealarcBP-83214
3993.000000-ISPY2 FatT2fseidealarcBP-87272
3994.000000-ISPY2 FieldmapHzT2fseidealarcBP-56951
3995.000000-ISPY2 IPT2fseidealarcBP-35774
3996.000000-ISPY2 OPT2fseidealarcBP-78000
50100.000000-ACRIN-6698 DWI TRACE from S5 bVals0100600800-11061
50200.000000-ACRIN-6698 ADC from S5 bVals0100600800-78462
59000.000000-ACRIN-6698 DWI MASK from S5 Whole Tumor Manual-51608
59001.000000-ACRIN-6698 DWI SEG from S5 Whole Tumor Manual-27228
61000.000000-ISPY2 VOLSER uni-lateral cropped SER-07197
61002.000000-ISPY2 VOLSER uni-lateral cropped PE2-27809
61006.000000-ISPY2 VOLSER uni-lateral cropped PE6-16627
61800.000000-ISPY2 VOLSER uni-lateral cropped original DCE-98836
61900.000000-ISPY2 VOLSER uni-lateral cropped Analysis Mask-08921

Posted by: XinYu @ Sept. 15, 2021, 7:46 a.m.

Looking into this. -Organizer-

Posted by: qin-organizer @ Sept. 16, 2021, 5:50 p.m.

Hi Xin Yu:

The naming of the derived series (all series with series number >20000) is described under the "Detailed Description" tag on the TCIA public wiki page at:
https://wiki.cancerimagingarchive.net/pages/viewpage.action?pageId=50135447#501354470d1c0b53772a4fba8580e8b19f65069b
The type of the series is coded into the last 4 digits of the series number; e.g., x0100 are all the DWI Trace images and x0200 are the 4 b-value ADC maps. See the descriptions on the wiki for all the codes.

I'm afraid that there is very wide variability in the naming of the original series, since each site in the multi-center trial has there own, and we did not mess with these other than adding the prefix "ISPY2:" or "ACRIN6698:" depending on whether the series is a standard ISPY2 protocol one (T2w, DCE, Scouts) or a 6698 specific one (4 b-value DWI).

SCOUT images typically have "Scout" or "Locator" in the description, and typically include images with different orientation in them. You can probably ignore these.

For DWI , if you are not familiar with multi-platform DWI conventions I highly recommend using the derived Trace images for analysis. But the series number for the original DWI is encoded in the description for the derived series: "ACRIN-6698 DWI MASK from S5" means the original was series 5. (On site used 3 2-b value scans, so in this case that would be series 5,6 and 7)

For DCE if you are only interested in the ipsilateral breast you can use the derived series, description e.g., "ISPY2 VOLSER uni-lateral cropped original DCE". If you want the original bi-lateral images you will need to look for either a single series with all phase images, or a set of similar series each with 1 phase volume in them. Typically the descriptions will have "DCE" or "Dyn" or "Ph<x>" in the series description.

Anything left after those identifications are probably the T2w series. These are the worst, as the study protocol just instructed sites to use whatever T2w sequence they preferred. The example you included in the Forum is a GE "IDEAL" sequence, which is a DIXON based sequence giving fat and water separate images. Hence all the series. You are probably safest taking the lowest non-scout series number for the T2w. In this case it is "3.000000-ISPY2 WATERT2fseidealarcBP", which is the water image from the IDEAL sequence.

I have been trying to find time to generate a guide document for the collection with the series type identifications. But it isn't trivial, and I haven't managed it yet. If I do get to it I will post it on the TCIA wiki and send out an email.

best regards,
David

Posted by: Newitt @ Sept. 22, 2021, 10:43 p.m.

Dear David,

Thank you very much for your help, it is very important to me.

Best regards,
XinYu

Posted by: XinYu @ Sept. 23, 2021, 6:27 a.m.
Post in this thread