Figure 8
The segmentation pipeline for ImageJ. Dense Golgi-stained particles are detected in the image using a simple thresholding operation (1). The resulting discrete objects are then joined into contiguous regions using a morphological closing operation (2). To smoothen these regions, a Gaussian blur filter (3) is applied, followed by a second thresholding operation (4). Non-specific staining is removed by sorting the segmented regions by area and discarding regions with area under the user-specified threshold (5). The remaining regions represent the result of segmentation and are added as a transparent channel to the original image stack (6). Alternatively, they can be returned as a stack of binary segmentation masks.
