Project Overview

DrawFAVA is a project whose goal is to annotate the boundaries of numerous functional regions of the human visual cortex. The human visual cortex is tiled with functional regions, most of which contain a map of the visual field. These regions can be identified in an individual human by first using expensive MR imaging then extensively processing the imaging data to estimate the visual field maps and finally manually tracing the boundaries based on the features of the maps. This process is expensive, difficult, and requires significant expertise.

Contemporary AI methods can be trained to automatically identify the boundaries of functional visual regions if they are given enough training data [Benson et al., 2025]. However, while there are several existing public datasets containing MR measurements of visual maps, few large datasets of annotations of the functional areas of those maps are publicly available. This project seeks to enable the training of AI methods to by generating a high quality dataset of visual area annotations across many datasets. The project further seeks to generate annotations from at least four separate researchers to examine reliability.

Datasets

The project aims to annotate the visual areas in seven public datasets:

• The Human Connectome Project 7 Tesla Retinotopy Dataset [Benson et al., 2018]

• The NYU Retinotopy Dataset [Himmelberg et al., 2021]

• The NYU-NEI Retinotopy Dataset (not yet published)

• The Natural Scenes Dataset [Allen et al., 2021]

• The CHN Retinotopic Mapping Dataset [Chang et al., 2025]

• The Forrest Gump Dataset [Hanke et al., 2014]

• The Natural Object Dataset [Gong et al., 2023]

Regions of Interest

The DrawFAVA project seeks to identify both retinotopic regions of interest (ROIs) in the early/posterior, dorsal, and ventral occipital cortex and category-selective regions in the ventral occipital temporal cortex (VOTC).

Retinotopic ROIs

Boundaries for the following retinotopic regions will be drawn on all subjects with retinotopic mapping data of sufficiently high quality. For more information on these ROIs and how to identify them, see the background chapter on the visual cortex.

• Posterior Occipital Cortex:

  • V1

  • V2

  • V3

• Ventral Occipital Cortex:

  • hV4

  • VO1

  • VO2

• Dorsal Occipital Cortex:

  • V3a

  • V3b

  • LO1

  • IPS0

Additionally, iso-eccentricity contours will be marked in the above regions for the following eccentricities:

• 0.5°

• 1°

• 1°

• 4°

• 7° (when possible)

• 11° (when possible)

Category Selective ROIs

Boundaries for the following category-selective regions in the VOTC will also be drawn on all subjects with sufficiently high quality data from the fLoc experiment [Stigliani et al., 2015].

• Faces

  1. OFA

  2. FFA-1

  3. FFA-2

  4. mTL-faces

  5. aTL-faces

• Bodies

  1. EBA

  2. FBA-1

  3. FBA-2

  4. mTL-bodies

• Places

  1. OPA

  2. PPA

  3. RSC

• Words

  1. OWFA

  2. VWFA-1

  3. VWFA-2

  4. mFS-words

  5. mTL-words