Fig. 1

MORPHIOUS workflow trains a one-class support vector machine to identify activated glial cells. A sliding window is applied to control (i.e., contralateral FUS, nonTG) hippocampal sections to extract morphological features (A). Extracted features are used to generate a spatial feature map (B). Selected morphological features from control hippocampal sections are used to train a one-class support vector machine which generates a decision boundary for defining non-activated microglia and astrocytes (C). A sliding window is further used to extract morphological features from test-sample (i.e., ipsilateral FUS, TgCRND8) hippocampal sections (D, E). The trained model is applied to test-sample hippocampal sections to identify outlier windows (F). Outliers are spatially clustered using the density-based spatial clustering of applications with noise algorithm (DBSCAN) to identify proximal clusters (G). To identify focal clusters, the integrated density of proximal cluster windows are sorted in ascending order, and the elbow point of this curve (red line) is used as a defined threshold value (H). DBSCAN is applied to windows with an integrated density above the defined threshold value to establish focal clusters (I).Contra., contralateral; FUS, focused ultrasound; Ipsi., ipsilateral; Hipp., hippocampus; Tg, TgCRND8 mice