Merge pull request #18 from brown-ccv/feat-binary-maps

[Feat] Binary map widget

README.md

@@ -71,7 +71,21 @@ You may pre-process the image with a Gaussian filter and contrast adjustments. D
 
 This process creates 2 additional layers (Keep track of these):
 - Pre-processed Data: An enhanced version of the dataset that sharpens cell borders and brightens the interior of blobs to help identify multiple cells within a single blob.
-- Cloud Binary: A binary image representing the region of interest with the highest probability of containing cells.
+
+### Create binary mask
+
+To improve the precision of our cell detection efforts within a specific channel, it is essential to create a binary mask. A binary mask is a black-and-white image that defines regions of interest where there is a higher likelihood of locating cells. By combining the segmentation results with this binary mask, we can effectively filter out noise generated during the segmentation process. This approach ensures that the analysis focuses exclusively on potential cells within the specified regions, improving both accuracy and efficiency in cell detection.
+
+ <img src="images/README/binary_mask_widget.png" alt="bimary mask widget" 
+   width="320" height="150">
+
+
+Use the dropdown menu to select the channel from which you want to create the binary mask. The threshold setting specifies that only pixels with an intensity greater than the defined percentage of the channel's maximum intensity will be included in the mask. This approach helps isolate regions with a higher likelihood of containing cells, enabling a more targeted and precise analysis by focusing on areas of interest while minimizing irrelevant data. Click on the "Create binary Mask" to add the result to the napari viewer
+
+
+ <img src="images/README/example_binary_mask.png" alt="Example binary mask" 
+   width="320" height="300">
+
 
 
 ### Data Segmentation

jaworski_cellseg/binary_maps_widget.py

@@ -0,0 +1,50 @@
+from magicgui import magicgui
+import napari
+import numpy as np
+from typing import Optional
+from skimage.morphology import remove_small_objects
+
+
+def create_binary_map_widget(
+    viewer: napari.Viewer,
+):
+    """
+    Widget to create biary map from the multiple channels
+    in the dataset
+    """
+
+    @magicgui(
+        call_button="Create Binary Mask",
+        auto_call=False,
+        channel_name={
+            "widget_type": "ComboBox",
+            "choices": [],
+        },
+        binary_map_threshold={"label": "Threshold", "min": 0.0, "max": 100.0},
+    )
+    def binary_map_widget(
+        channel_name: Optional[str] = None,
+        binary_map_threshold: float = 90.0,
+    ):
+        cloud_channel = viewer.layers[channel_name].data
+        cloud_percentile_float = np.percentile(
+            cloud_channel.astype(np.float64), binary_map_threshold
+        )
+
+        # Create a binary mask using the percentile as a threshold
+        binary_mask = cloud_channel > cloud_percentile_float
+        binary_mask = remove_small_objects(binary_mask, min_size=100)
+        viewer.add_image(
+            binary_mask, name=f"{channel_name} binary mask perc={binary_map_threshold}"
+        )
+
+    def update_channel_names_dropdown(event=None):
+        channel_layer_names = []
+        for layer in viewer.layers:
+            if "Channel" in layer.name:
+                channel_layer_names.append(layer.name)
+
+        binary_map_widget.channel_name.choices = channel_layer_names
+
+    viewer.layers.events.inserted.connect(update_channel_names_dropdown)
+    return binary_map_widget

jaworski_cellseg/jwski_widget.py

@@ -7,6 +7,7 @@
 from .label_counting_widget import create_label_counting_widget
 from .subregion_selection_widget import create_region_selection_widget
 from .configuration_widget import create_configuration_widget
+from .binary_maps_widget import create_binary_map_widget
 import yaml
 from pathlib import Path
 
@@ -44,6 +45,7 @@ def init_ui(self):
         self.pre_process_data_widget = create_pre_process_data_widget(
             self.viewer, self.inferer_widget, self.current_config
         )
+        self.binary_map_widget = create_binary_map_widget(self.viewer)
         self.count_widget = create_label_counting_widget(
             self.viewer, self.physical_sizes
         )
@@ -56,6 +58,7 @@ def init_ui(self):
             "config_widget",
             "data_widget",
             "pre_process_data_widget",
+            "binary_map_widget",
             "count_widget",
             "region_selection_widget",
             "inferer_widget",

jaworski_cellseg/label_counting_widget.py

@@ -86,7 +86,9 @@ def cell_counting_widget(
         count = len(filtered_labels_regions)
         cell_counting_widget.total_count.value = str(count)
 
-        viewer.add_labels(result_image, name=f"{count} Filtered Labled layer", opacity=1.0)
+        viewer.add_labels(
+            result_image, name=f"{count} Filtered Labled layer", opacity=1.0
+        )
 
         file_path = cell_counting_widget.save_results.value
         if file_path:
@@ -113,7 +115,7 @@ def update_layer_dropdown(event=None):
 
         # Get names of binary masks layers in the viewer
         binary_masks_layer_names = [
-            layer.name for layer in viewer.layers if "cloud binary" in layer.name
+            layer.name for layer in viewer.layers if "binary mask" in layer.name
         ]
         # Update the dropdown choices for label_layer_name
         cell_counting_widget.binary_layer_name.choices = binary_masks_layer_names

jaworski_cellseg/pre_process_data_widget.py

@@ -16,7 +16,6 @@ def pre_process_bio_data(
     contrast_adjustment,
     contrast_lower,
     contrast_upper,
-    binary_map_threshold,
 ):
     try:
         channel_data = napari_viewer.layers["Channel 0"].data
@@ -37,15 +36,8 @@ def pre_process_bio_data(
             print(f"Applying Gaussian fileter sigma {gaussian_factor}")
             img_rescale = gaussian_filter(img_rescale, sigma=gaussian_factor)
 
-        cloud_channel = napari_viewer.layers["Channel 2"].data
-        cloud_percentile_float = np.percentile(
-            cloud_channel.astype(np.float64), binary_map_threshold
-        )
+        return img_rescale
 
-        # Create a binary mask using the percentile as a threshold
-        binary_mask = cloud_channel > cloud_percentile_float
-        binary_mask = remove_small_objects(binary_mask, min_size=100)
-        return img_rescale, binary_mask
     except Exception as e:
         print(e)
 
@@ -61,28 +53,22 @@ def create_pre_process_data_widget(
         contrast_adjustment={"label": "Contrast Adjustment (Percentiles)"},
         contrast_lower={"label": "min", "min": 0.0, "max": 100.0},
         contrast_upper={"label": "max", "min": 0.0, "max": 100.0},
-        binary_map_threshold={"label": "Binary Threshold", "min": 0.0, "max": 100.0},
     )
     def pre_process_data_widget(
         gaussian_checkbox: bool = config.get("gausianFilter", True),
         gaussian_factor: float = config.get("gaussian_sigma", 1.0),
         contrast_adjustment: bool = config.get("contrast_adjustment", True),
         contrast_lower: float = config.get("contrast_min", 5.0),
         contrast_upper: float = config.get("contrast_max", 95.0),
-        binary_map_threshold: float = config.get("binary_map_threshold", 90.0),
     ) -> "napari.types.LabelsData":
 
-        result_image, binary_mask = pre_process_bio_data(
+        result_image = pre_process_bio_data(
             viewer,
             gaussian_checkbox,
             gaussian_factor,
             contrast_adjustment,
             contrast_lower,
             contrast_upper,
-            binary_map_threshold,
-        )
-        viewer.add_image(
-            binary_mask, name=f"cloud binary mask perc={binary_map_threshold}"
         )
         viewer.add_image(result_image, name="Pre processed data")
         ## In case more layers are added, lets set the Tiff layer by default

jaworski_cellseg/subregion_selection_widget.py

@@ -45,7 +45,7 @@ def region_selection_widget(
 
         modified_data = binary_layer.data * mask_3d
 
-        viewer.add_image(modified_data, name="cloud binary region")
+        viewer.add_image(modified_data, name=f"{shape_layer_name} binary region")
 
     # Update dropdown menu options in the `region_selection_widget` to display
     # available shape and binary layers in the Napari viewer.
@@ -56,7 +56,7 @@ def update_binary_layers_dropdown(event=None):
         for layer in viewer.layers:
             if isinstance(layer, Shapes):
                 shape_layer_names.append(layer.name)
-            if "cloud binary" in layer.name:
+            if "binary mask" in layer.name:
                 binary_layer_names.append(layer.name)
 
         # Update the dropdown choices for label_layer_name