Georgia State University
Loud, Red Trucks
My work, as a painter, often focuses on everyday objects grouped in a way that seems to tell a story about the people who brought them all together. The way I set up my still life objects is just as important as how I paint them. When I talked to Thomas about his work, we realized we could implement the analysis techniques he uses to identify signatures in network brain activity on a series of still life paintings. In essence, we would be using the programs to identify “artistic signatures” in my paintings, perhaps visual and compositional similarities that I tend to repeat.
We analyzed four paintings total: three older paintings (figures 1-3) and one new one made specially for this project (above). The painting Loud, Red Trucks was inspired by another facet of Thomas’ work. Passionate about autism research, Thomas conducted his thesis work on identifying signature differences in striatal resting states in brains of those with and without autism. One of the characteristics of autism is repetitive interest or fixation on certain subject matter or special objects, such as trains, cameras, stuffed animals, batteries, etc. Because his research focus was largely analyzing brains of those with autism, I decided to organize this still life around some of the special objects that fascinate children, especially those with autism. Models of emergency vehicles are arranged neatly over an array of flags of the world, and planes dart in and out through the composition as a Lego man in his own homemade airplane teeter over a stack of books. Cards, batteries, shoelaces and various cords loop throughout, and the pile is topped with a bouquet-like display of thermometers – which are objects of fixation for all of us during this pandemic.
Farmer's Market, 2017
Polka Dots, 2018
Our work - the paintings, analysis, and visual outputs - all taken together, are a playful meditation on the limits of traditional neuroimaging pipelines. Thomas, in his postdoctoral work, has turned to exploring multi-model data fusion, combining multiple neuroimaging methods in order to better understand large-scale network activity in the brain.
The images below displayed are components resolved using Source-based Morphometry (SBM) of 2D images. SBM utilizes a technique called principal component analysis (PCA) to identify which “parts” of an image or set of images account for the greatest variance across one or all of the images. This is done as a data reduction step (e.g. 14 images to 4 principal components).
Figure 5 Component 1
Figure 7 Component 3
Figure 6 Component 2
Figure 8 Component 4
Next, a technique called Independent Component Analysis (ICA) is used on the principal components to break them down into independent regions with the most similar patterns. These form the “Independent components” of the artwork (in greyscale Figures 5-8). These components, in red and blue, have also been overlaid on Loud, Red Trucks to allow the viewer to explore how the patterns may or may not be repeated in the final painting.
Figure 9 ICA 1 overlay
Figure 11 ICA 3 overlay
Figure 10 ICA 2 overlay
Figure 12 ICA 4 overlay
Finally, the multi-colored 4x4 matrix is the correlation between each independent component, with warm colors showing greater similarity between two independent components, and cool values indicating on component decrease as another increases.