Skip to content

Repository for the Bloch-McConnell simulation (BMsim) challenge

Notifications You must be signed in to change notification settings

pulseq-cest/BMsim_challenge

Repository files navigation

== BMsim challenge ==

Welcome to the repository of the Bloch-McConnell simulation (BMsim) study / challenge. The idea of the project can be summarized as follows:

  1. Every participant simulates different well-defined cases / scenarios
  2. The simulation results from all participants are collected online
  3. An online evaluation script enables a live comparison and validation against all other results

In the first study, we have chosen 3 different preparation schemes consisting of single block/CW pulses and 2 different pool systems resultung in 4 different simulation cases (cases 1 - 4). More infos and a summary of our findings and achievements can be found in the abstract and slides from the ISMRM 2023.

The second study (cases 5 - 8) covers more complex cases including shaped pulses and pulse trains. An abstract submission for the ISMRM 2024 is planned.

Simulation results

To keep the burden for submitting your simulation results as low as possible, we decided to collect the results in a simple Google Docs spreadsheet.

Please feel free to add your name / group in case it's not listed yet and post your results.

Online evaluation script

To provide a simple way to compare your own simulation results with (a selection of) the results collected in the Google Docs spreadsheet, we set up an online evaluation script in form of a iPython Notebook that is hosted on Google colab.

Simulation cases

General settings / assumptions

  1. fully relaxed initial magnetization (Zi = 1) for every offset (this is equivalent to a very long recovery time t > 10 * T1)
  2. post-preparation delay = 6.5 ms (in the pulseq-file this corresponds to the gradient spoiler duration)
  3. gyromagnetic ratio: 42.5764 MHz/T (see FAQ below)
  4. larmor frequency (3T): 127.7292 MHz/T
  5. Normalization scan at -300 ppm

Second study (cases 5 - 8)

Case 5: 2 pool model, single shaped pulse

  • pool model: 2 pool model of creatine as defined in case_5_2pool_model.yaml
  • prep. details:
    • pulse shape: Gaussian
    • pulse duration: 50 ms
    • number of pulses: 1
    • total saturation time: 50 ms
    • pulse power (B1rms): 1.9962 µT
    • offset list: -2:0.02:2 ppm

More details about the pool model and preparation scheme can be found in the corresponding README

Case 6: 2 pool model, pulsed APTw preparation

  • pool model: 2 pool model of creatine as defined in case_6_2pool_model.yaml
  • prep. details:
    • pulse shape: Gaussian
    • pulse duration: 50 ms
    • number of pulses: 36
    • interpulse delay: 5 ms
    • number of interpulse delays: 35
    • total saturation time: 1.975 s
    • pulse power (B1rms): 1.9962 µT
    • offset list: -15:0.1:15 ppm

More details about the pool model and preparation scheme can be found in the corresponding README

Case 7: 5 pool model, pulsed APTw preparation

  • pool model: 5 pool model of WM as defined in case_7_5pool_model.yaml
  • prep. details:
    • pulse shape: Gaussian
    • pulse duration: 50 ms
    • number of pulses: 36
    • interpulse delay: 5 ms
    • number of interpulse delays: 35
    • total saturation time: 1.975 s
    • pulse power (B1rms): 1.9962 µT
    • offset list: -15:0.1:15 ppm

More details about the pool model and preparation scheme can be found in the corresponding README

Case 8: 5 pool model, (modified) WASABI preparation

  • pool model: 5 pool model of WM as defined in case_8_5pool_model.yaml
  • prep. details:
    • pulse shape: block
    • pulse duration: 5 ms
    • number of pulses: 2
    • interpulse delay: 100 µs
    • number of interpulse delays: 1
    • total saturation time: 0.0101 s
    • pulse power (peak): 3.7 µT
    • offset list: -2:0.05:2 ppm

More details about the pool model and preparation scheme can be found in the corresponding README

First study (cases 1 - 4)

Case 1: 2 pool model, APTw preparation - steady-state

  • pool model: 2 pool model of creatine as defined in case_1_2pool_model.yaml
  • prep. details:
    • pulse shape: block
    • pulse duration: 15 s
    • pulse power: 2 µT
    • offset list: -15:0.1:15 ppm

More details about the pool model and preparation scheme can be found in the corresponding README

Case 2: 2 pool model, APTw preparation

  • pool model: 2 pool model of creatine as defined in case_2_2pool_model.yaml
  • prep. details:
    • pulse shape: block
    • pulse duration: 2 s
    • pulse power: 2 µT
    • offset list: -15:0.1:15 ppm

More details about the pool model and preparation scheme can be found in the corresponding README

Case 3: 5 pool model, APTw preparation

  • pool model: 5 pool model of WM as defined in case_3_5pool_model.yaml
  • prep. details:
    • pulse shape: block
    • pulse duration: 2 s
    • pulse power: 2 µT
    • offset list: -15:0.1:15 ppm

More details about the pool model and preparation scheme can be found in the corresponding README

Case 4: 5 pool model, WASABI preparation

  • pool model: 5 pool model of WM as defined in case_4_5pool_model.yaml
  • prep. details:
    • pulse shape: block
    • pulse duration: 5 ms
    • pulse power: 3.7 µT
    • offset list: -2:0.05:2 ppm

More details about the pool model and preparation scheme can be found in the corresponding README

FAQ

  • How did you choose the value of the gyromagnetic ratio?

    The NIST value of the shielded proton gyromagnetic ratio is 2.675153151 x 108 s-1 T-1. Dividing this value by 2 Pi yields 42.576384750950949004433240733872 MHz/T, which results in the used value of 42.5764 MHz/T when rounded to 4 digits.

    Please make sure to use these values for gamma in your simulations:

    • 42.5764 MHz/T
    • 42.5764 x 2 x Pi s-1 T-1 (do NOT use the exact NIST value)
  • How do you define the pool size fraction f?

    There are two different options to define the pool size fractions:

    1. define water f=1, and all other fractions relative to water
    2. define M0_i of each pool i and then normalize f_i= M0_i/sum(M0_i)

    The first study showed that all groups participating so far use definition 1. Therefore, we encourage all new participants to use this definition as well.

  • How do you define the MT pool?

    Some simulations use x, y, and z components to describe a Lorentzian MT pool. Others use only the z-component and assume a Lorentzian lineshape factor there. The results from the first study showed that these different implementations are NOT INTERCHANGEABE right now. Therefore, we suggest treating the MT pool similar to a CEST pool and consider all 3 components.

About

Repository for the Bloch-McConnell simulation (BMsim) challenge

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published