Optimization Strategies in Connective Field Mapping
##################################################################

Benchmarks for different modelling procedures in cortical connective field (CF) mapping.
========================================================================================



Here we compare three flavors of population cortico-cortical connective field (CF) modeling. First, the former implementation in ``mrVista`` (using Matlab's ``lscov``) :footcite:p:`Haak_2013,Gravel_2014`, which implemented a grid search approach. Without further optimization, the parameter space revealed by this implementation match predefined grid value predictions. Second, the implementation part of the *connective field* branch of the `prfpy <https://github.com/VU-Cog-Sci/prfpy>`_ Python package :footcite:p:`Knapen_topo_2021`. Using Python scikit-learn optimization functions (implemented as part of the ``prfpy`` package), the parameter space ``prfpy`` provides also converge to predefined grid, although at a much faster rate (thanks to the CPU parallelization tool `joblib``). Third, a custom version of more recent CF modelling approach (by So-Hyeon et al. (2024) :footcite:p:`Yoo_proto_2024`) that implements a *derivative-free* parameter (CF size) refinement approach (`see here <https://nicogravel.github.io/cfMaps/html/cfmap.html#cfmap.CF_mapping.optimize_connfield_dfree>`_). Fourth, our custom Python implementation of *automatic differentiation* -powered `gradient descent <https://en.wikipedia.org/wiki/Gradient_descent>`_ using ``TensorFlow`` and CUDA (partly inspired in the recent Python package `braincoder <https://braincoder-devs.github.io/>`_) :footcite:p:`deHollander2024braincoder` to achieve highly efficient gradient descent (`see here <https://nicogravel.github.io/cfMaps/html/cfmap.html#cfmap.CF_mapping.optimize_connfield_gdescent>`_ and `here <https://nicogravel.github.io/cfMaps/html/cfmap.html#cfmap.CF_mapping.optimize_connfield_joint>`_ for a joint optimization approach that optmimzed both CF size and position). To coordinate this effort, we rely on widely used population receptive field mapping tools :footcite:p:`Dumoulin_2008,Benson_2018` and high-field 7T-MRI retinotopy data kindly made available by `NeuroSpin <https://joliot.cea.fr/drf/joliot/en/Pages/research_entities/NeuroSpin.aspx>`_.



How to cite
***********

If you use this code please cite using the following information:

Gravel, N., Zhan M., Renken, R., & Cornelissen, F. W. (2025). *Optimization strategies in cortical connective field  mapping*. Zenodo. DOI: https://doi.org/10.5281/zenodo.17373320

Alternatively, use this BibTeX entry:

.. code-block:: bibtex

   @misc{Gravel_optCF_2025,
     author       = {Gravel, Nicol{\'a}s and Zhan, Minye and Renken, Remco and Cornelissen, Frans W},
     title        = {Optimization strategies in cortical connective field mapping},
     year         = {2025},
     publisher    = {Zenodo},
     doi          = {10.5281/zenodo.17373320},
     url          = {https://doi.org/10.5281/zenodo.17373320}
   }




**********
Content
**********
.. toctree::
      :maxdepth: 2
      :numbered:          

      Visual Field mapping <vfm/index>
      Connectivity mapping <connmaps/index>
      References <refs>
      Package <modules>  

.. note::

	This online resource is an evolving document. It may therefore contain errors and corrections. Content will be added over time. Check it regularly to get the latest version.

**********
References
**********

.. footbibliography::