2. Workflow
Figure 2. GUI allows user to set parameters for tasks implemented with functions and classes contained in MWSegEval libraries.
Figure 3. Work flow for segmentation and image analysis
The complete set of tasks and order that they are carried out is shown in figure 3. Each task is selected from the task list presented by the GUI shown in figure 2. The results generated by each task are displayed for the user to review before continuing to the next task. Each task may be repeated multiple times with different valued user parameters.
For example, when the thresholding technique is selected (i.e., the automatic segmentation option is not selected), then the user may choose to run the segmentation task and subsequent analysis tasks again with different threshold levels. This allows the user to observe the masks and metrics that are generated and experiment with different threshold values used to classify different tissue types, without running through the entire workflow. An example of this approach is demonstrated in Demonstration 4 – Extremely dense breast – thresholding segmentation technique of section 3 in which the thresholding technique is applied to a microwave image of an extremely dense classified breast using various threshold values.
Likewise, the user may run the segmentation task multiple times using different values for the Contour extraction threshold% to observe the impact that adjustments of this parameter has on filtering spurious artifacts that may form along interfaces during image reconstruction. An example demonstrating the impact that adjustments of the Contour extraction threshold% has on the presence of artifacts in the tissue masks and tissue type maps is described in section 2.2C Tissue artefact filtering.
Each task is typically run in the sequence that they appear in the list, as each task uses the results of the preceding task.