Mapping the "What" and "Where" Visual Cortices and Their Atrophy in Alzheimer's Disease: Combined Activation Likelihood Estimation with Voxel-Based Morphometry

Abstract

The human cortical regions for processing high-level visual (HLV) functions of different categories remain ambiguous, especially in terms of their conjunctions and specifications. Moreover, the neurobiology of declined HLV functions in patients with Alzheimer's disease (AD) has not been fully investigated. This study provides a functionally sorted overview of HLV cortices for processing "what" and "where" visual perceptions and it investigates their atrophy in AD and MCI patients. Based upon activation likelihood estimation (ALE), brain regions responsible for processing five categories of visual perceptions included in "what" and "where" visions (i.e., object, face, word, motion, and spatial visions) were analyzed, and subsequent contrast analyses were performed to show regions with conjunctive and specific activations for processing these visual functions. Next, based on the resulting ALE maps, the atrophy of HLV cortices in AD and MCI patients was evaluated using voxel-based morphometry. Our ALE results showed brain regions for processing visual perception across the five categories, as well as areas of conjunction and specification. Our comparisons of gray matter (GM) volume demonstrated atrophy of three "where" visual cortices in late MCI group and extensive atrophy of HLV cortices (25 regions in both "what" and "where" visual cortices) in AD group. In addition, the GM volume of atrophied visual cortices in AD and MCI subjects was found to be correlated to the deterioration of overall cognitive status and to the cognitive performances related to memory, execution, and object recognition functions. In summary, these findings may add to our understanding of HLV network organization and of the evolution of visual perceptual dysfunction in AD as the disease progresses.

Keywords: Alzheimer's disease; activation likelihood estimation; functional magnetic resonance imaging; visual perception; voxel-based morphometry.

Figure 1

Activation likelihood estimation (ALE) maps of “what” and “where” visions (p < 0.01, false discovery rate corrected). (A) Left panel is the axial view of the ALE activation map for face vision. Lower right 3D image and upper coronal image show the significant clusters (in orange) in bilateral occipital lobe and right superior temporal sulcus. (B) Left panel is the axial view of the ALE activation map of word vision. Upper right 3D image and lower sagittal image show the significant clusters in the left inferior and middle temporal gyrus and the inferior frontal gyrus. (C) Left panel is the axial view of the ALE map for object vision. The magnified 3D and sagittal images display the significant clusters in the bilateral occipital lobe, right fusiform gyrus and inferior frontal gyrus. (D) Left panel is the axial view of the ALE activation map for spatial vision. Upper right 3D and lower sagittal images show the significant clusters in the right superior occipital gyrus and the right superior parietal lobule. (E) Left panel is the axial view of the ALE activation map for motion vision. The 3D images in the right panel display the significant clusters located in the bilateral tempo-occipital regions, right superior parietal gyrus, bilateral postcentral gyrus and left supramarginal gyrus.

Figure 2

Specific regions of high-level visions from contrast analyses (p < 0.01, false discovery rate corrected). (A) Face-specific regions are obtained from the comparisons of face vs. word (orange) and face vs. object (blue). (B) Word-specific regions are obtained from contrast analyses of word vs. face (orange) and word vs. object (blue). (C) Clusters in orange indicate a higher activation of the “what” vision than “where” vision, while clusters in blue show a higher activation of the “where” vision.

Figure 3

Conjunctive regions of high-level visions (p < 0.01, false discovery rate corrected). The axial images in the left panel show the overview of the conjunctive regions for high-level visions of what & where (orange), face & word (green), face & object (pink), and word & object (blue). The magnified 3D image and the sagittal images in the right panel display that most of the conjunctive regions are convergently distributed in the bilateral lateral occipital complex (LOC).

Figure 4

Significantly atrophied high-level visual cortices in patients with Alzheimer's disease (AD) and mild cognitive impairment (MCI). The figure shows high-level visual regions of interest (ROIs) with significantly reduced mean gray matter volume in AD and late MCI (LMCI) groups. These ROIs are respectively corresponding to the face (A), word (B), object (C), motion (D), and spatial (E) visual processing. Regions in green color are found significantly atrophied only in AD group, while regions in blue are found significantly atrophied in both AD and LMCI groups.