Skip to main content
European Commission logo
English English
CORDIS - EU research results
CORDIS
CORDIS Web 30th anniversary CORDIS Web 30th anniversary

Multimodal characterization of the visual word form area: An integrative computational model

Periodic Reporting for period 2 - ReCiModel (Multimodal characterization of the visual word form area: An integrative computational model)

Reporting period: 2020-09-01 to 2021-08-31

The ventral occipito-temporal (vOT) association cortex contributes significantly to recognize different types of visual patterns. It is widely accepted that a subset of this circuitry, including the visual word form area (VWFA), becomes trained to identify word forms. Nevertheless, due to heterogeneous experimental procedures across studies and intrinsic limitations of functional and structural MRI tools, the exact cortical location of what it is referred as the VWFA typically differs between studies. Additionally, it should be expected that different adjacent brain tissue within the vOT to perform different specific computations. I proposed to conduct the first systematic investigation combining functional and structural MRI data to further characterize spatially segregated VWFAs. Analogously, the functional and structural connectivity between these VWFAs and other relevant visual and language brain regions was examined as well. In the last year of the grant, covered by this report, the objective was to incorporate the data into a detailed statistical model intended to predict reading behavior. The ultimate goal of the project is to create a highly detailed characterization of the early stages of reading and to establish a baseline model and parameter range that will serve to clarify differences between typical and atypical readers.
I was able to finish the main objectives set up for the period and to validate the hypotheses delineated in the project. Those results were published in a high impact journal (PNAS), and they are being highly cited (57 at the moment of writing this report). I am still working with this dataset, for example:

• I repeated the analyses in the right hemisphere, and I just finished writing the manuscript.
• I am performing a highly detailed work in individual space as well, trying to quantify how well these results hold at the individual subject level. This is still ongoing.
• The resting state and functional localizer data is being used in a project to compare two different types of functional connectivity. This work is in the writing phase.
• I directed a Master Thesis and we used this dataset to analyse the structural connectivity of the VOTC with the IFG. The Master Thesis was defended in July 2021, paper is almost finished as well.
• This project led to some past and ongoing collaborations as well. One of those collaborations was already published in Nature Communications.

I was able to work in state-of-the art neuroimaging tools and procedures to accomplish the last objective of the project: creating a model and baseline of healthy readers. In order to create models that can be used across scanners and be useful outside the research environment, I started working on quantitative MRI methods and the translation of research findings to the clinic, trying to find ways of identifying differences at the individual subject level. I developed a conceptual and experimental project to understand the replication and generalization in applied neuroimaging, and the result of that work was published in the journal NeuroImage. In order to have a platform to validate the results of our analyses, I worked in a software validation platform because I wanted to apply it to a quantitative fMRI models (population receptive fields -pRF-). This work was recently published in PLOS Computational Biology.
Regretfully, the fourth objective is still ongoing. The results will need to wait at least one year. I was relying on BCBL’s data acquisition efforts with a large dyslexic cohort. This work has been significantly delayed mostly due to COVID. Nevertheless, I was able to do other significant work with the methods I developed. I applied one of the methods to show that the receptive fields in early visual cortex are nearly circular (published early 2021 at Journal of Neuroscience). Additionally, I’ve been exploring other datasets that could help with the modelling and differentiation of the dyslexic population. In one very exciting result, we help interpret sensory and cognitive signals in visual and reading areas. This work has been posted to a preprint server (bioRxiv) and is under review in the journal Nature Communications.
All the results went beyond the state of the art and provided incremental contributions to the field. Summary of the 3 main contributions: the work in VWFA published in PNAS set a framework for other works that used our results to build upon them (White et al. 2019 PNAS, or Yablonski et al., 2021, presented at the Annual Conference of the Society for the Neurobiology of Language). The NeuroImage paper proposed a framework for replication and generalization, and in top of that proposed a novel method to increase the precision of the metrics we use. Similar works have been published recently, as in Chamberland et al., (2021; Nature Computational Science). The Plos Comp. Biol. paper on software validation provided a novel and reproducible framework to check the validity of neuroimaging algorithm results, and in top of it, extended what we knew about the pRFs by discovering a dependence of the estimated size of the pRF on the HRF assumed by the model. We provided recommendations and improvements in how to minimize these dependencies. These findings will allow to revisit the literature with new eyes and will help design new experiment and new generation analysis tools. We already used the technology to show how 1 published result was wrong, most probably because of an error on the software they used. This work was published in Journal of Neuroscience. In my previous report I was hopeful to overcome the COVID related delays and that the BCBL would be able to scan the participants with dyslexia that were part of my third-year analyses. I was hoping to apply a novel computational method to try to rank individual dyslexic participants versus the baseline of the healthy readers in different metrics. We were not able to scan the dyslexic participants on time, so I continued working with the detailed baseline model and existing subjects in BCBL’s and Stanford’s databases. Using one database from Stanford, I was able to provide a new insight in how we use sensory and cognitive signals in the visual and reading ventro-occipital reading regions (this work is still under review in Nature Communications). I think that the impact of the previous results will be building over time. My plan is to continue focusing on using behavioural, functional and structural Magnetic Resonance Imaging techniques to investigate the neural basis of vision and reading and developing functional and structural MRI computational methods to further examine cognitive functions and enhance neuroimaging reproducibility, validity and generalizability. My long-term career objective is to develop a clinical magnetic resonance imaging diagnostic tool to help those struggling to read. Developmental dyslexia is the most prevalent reading disability in the population (3–7% depending on definitional criteria and language orthography), with its manifestations ranging from specific inabilities to decode words to higher level language limitations. The educational, social and economic impact on the individual can be life-altering. The diagnostic tool needs to be at the individual level and applicable in any standard clinic with a MRI machine. The impact of such a tool can be enormous in those suffering the disabilities in particular and the society in general.
figure1.jpg