TF6.3 — IVIM algorithm comparison¶
The taskforce aims to study, quantify and explain the differences between IVIM fitting algorithms using the OSIPI IVIM code collection. With simulated and in-vivo data, systematic and random errors will be quantified and the influence of harmonization of algorithm hyper parameters will be studied. With this the impact of IVIM fitting algorithms on reproducibility is shown and possible recommendations on the use of fitting algorithms can be made.
Leads¶
Ongoing projects¶
The initial results from the unit tests of the code collection show that there is a large influence of the used fitting algorithm on IVIM parameter calculations, which contributes to the poor reproducibility in literature. In addition, many centers use in-house developed algorithms making it difficult to reproduce results. In the current project, we are performing an in depth analysis of all algorithms to study how large the influence of fitting algorithms is on the calculated IVIM parameters. With these results we can estimate the reproducibility of IVIM parameter calculations and propose recommendations on analysis details, like fitting bounds and initial guesses.
More info can be found in the roadmap