use a class to manage the nodes involved in a slicer plot view

slicer_plot_data.py

# To try this example, save this to a file slicer_plot_data.py, then execute Slicer from the command line
# with the arguments `--python-script slicer_plot_data.py`.

import slicer, qt, vtk

PLOT_TYPES = {
  "line" : slicer.vtkMRMLPlotSeriesNode.PlotTypeLine,
  "bar" : slicer.vtkMRMLPlotSeriesNode.PlotTypeBar,
  "scatter" : slicer.vtkMRMLPlotSeriesNode.PlotTypeScatter,
  "scatterbar" : slicer.vtkMRMLPlotSeriesNode.PlotTypeScatterBar,
}

class SlicerPlotData:
  """Container for and manager of the nodes associated to a slicer plot view."""

  # A plot view node we will keep empty in order to have a way of displaying no plot
  empty_plot_view_node = slicer.mrmlScene.AddNewNodeByClass("vtkMRMLPlotViewNode", "SlicerPlotDataEmptyPlotViewNode")

  def __init__(self, name:str):
    """Create SlicerPlotData, making a qMRMLPlotView and a vtkMRMLPlotViewNode with the given name."""
    self.name = name
    self.plot_view = slicer.qMRMLPlotView()
    self.plot_view.setMRMLScene(slicer.mrmlScene)
    self.plot_view_node = slicer.mrmlScene.AddNewNodeByClass("vtkMRMLPlotViewNode", name+"PlotView")
    self.plot_view.setMRMLPlotViewNode(self.plot_view_node)
    self.plot_nodes = {} # chart, table, and series; see the parameter "nodes" in the doc of slicer.util.plot


  def set_plot_data(self, data, x_axis_label=None, y_axis_label=None, title=None, legend_label=None, plot_type="line", labels=None):
    """
    Populate the plot with the data from the given numpy array.

    Args:
      data: a numpy array of shape (N,2) containing the data to plot
      x_axis_label: the title of the x-axis to display
      y_axis_label: the title of the y-axis to display
      title: plot title; also shows up in the names of helper nodes
      legend_label: the text to put in the legend
      plot_type: one of "line", "bar", "scatter", or "scatterbar"
      labels: a list of string labels-- this affects bar and scatterbar plot types
    """

    if title is None: title = self.name
    if legend_label is None: legend_label = title

    if len(data.shape)!=2 or data.shape[1]!=2:
      raise ValueError(f"data was expected to be a numpy array of shape (N,2), got {tuple(data.shape)}")

    # Here we avoid changing plots while they are associated to a plot view.
    # This is to supress an error that otherwise shows up
    # (see e.g. https://github.com/KitwareMedical/lungair-desktop-application/issues/27).
    self.plot_view.setMRMLPlotViewNode(self.empty_plot_view_node)

    if x_axis_label is not None and y_axis_label is not None:
      columnNames = [x_axis_label, y_axis_label]
    else:
      columnNames = None

    plot_chart_node = slicer.util.plot(
      data, 0, show = False,
      title = title,
      columnNames = columnNames,
      nodes = self.plot_nodes
    )
    plot_chart_node.SetXAxisTitle(x_axis_label)
    if y_axis_label is not None:
      plot_chart_node.SetYAxisTitle(y_axis_label)
    assert(len(self.plot_nodes["series"]) == 1)
    self.plot_nodes["series"][0].SetName(legend_label) # This text is displayed in the legend
    self.plot_nodes["series"][0].SetPlotType(PLOT_TYPES[plot_type])
    self.plot_view_node.SetPlotChartNodeID(plot_chart_node.GetID())

    if labels is not None:
      labels_array = vtk.vtkStringArray()
      for label in labels:
        labels_array.InsertNextValue(label)
      label_column_name = (x_axis_label if x_axis_label else "X-axis") + " Label"
      labels_array.SetName(label_column_name)
      self.plot_nodes['table'].AddColumn(labels_array)
      self.plot_nodes["series"][0].SetLabelColumnName(label_column_name)

    self.plot_view.setMRMLPlotViewNode(self.plot_view_node)


#------------------------------------------
# Example usage
#------------------------------------------

import numpy as np
slicer_plot_data = SlicerPlotData("example plot")
slicer_plot_data.set_plot_data(np.random.normal(size=(10,2)))

# Now you can put the widget slicer_plot_data.plot_view wherever you want. Here we just show it on its own:
slicer_plot_data.plot_view.show()

# (Note that for proper cleanup when exiting the application, the plot_view widget should really be parented to something.)

# Use set_plot_data to update what is shown in the plot view:
slicer_plot_data.set_plot_data(np.random.normal(size=(10,2)))