Figure 3-. Evaluation of covariance within two different regions of a reconstructed object.
Focused classifications using 2-D masks, applied to an arm/hand region (to the right of body in figure and encompassing both arm/hand conformations), were performed using either (A-B) the dataset for the 2nd scenario or (C-D) the dataset for the 3rd scenario, both at an SNR of 0.100. In all cases, four classes were recovered for the different asymmetric units (arms in two positions, each with and without a hand), and are displayed in panels A and C. (A) Volumes recovered from focused classification in the 2nd scenario, where all components are randomly occupied (control). (B) Correlation matrix describing the relationships between the near hand, near foot, and far foot. (C) Volumes recovered from focused classification in the 3rd scenario, where the near hand is always co-associated with the near foot. (D) Correlation matrix describing the relationships between the components. Near hand, where focused classification is performed, is circled in blue, near foot is circled in purple, and far foot is circled in pink. In the tables, the values are colored using a gradient: −1 (green, anti-correlated) < 0 (red, not correlated) < 1 (green, correlated).