Register CLEM PNGs (#719)

* Added new fields to the Pydantic models for the CLEM results to register thumbnail images for ISPyB with
* Added comments to 'GridSquareParameters' table to clarify what the different fields are used for
* Added new columns to the 'CLEMImageSeries' database table to store information about the thumbnails used when registering data to ISPyB
* Updated DataCollectionGroup and GridSquare registration logic for CLEM workflow to make use of scaled down thumbnails instead of full-sized TIFFs

Author: tieneupin

src/murfey/util/db.py

@@ -239,7 +239,8 @@ class CLEMImageSeries(SQLModel, table=True):  # type: ignore
     series_name: str = Field(
         index=True
     )  # Name of the series, as determined from the metadata
-    search_string: Optional[str] = Field(default=None)  # Path for globbing with
+    image_search_string: Optional[str] = Field(default=None)
+    thumbnail_search_string: Optional[str] = Field(default=None)
 
     session: Optional["Session"] = Relationship(
         back_populates="image_series"
@@ -295,9 +296,12 @@ class CLEMImageSeries(SQLModel, table=True):  # type: ignore
     number_of_members: Optional[int] = Field(default=None)
 
     # Shape and resolution information
-    pixels_x: Optional[int] = Field(default=None)
-    pixels_y: Optional[int] = Field(default=None)
-    pixel_size: Optional[float] = Field(default=None)
+    image_pixels_x: Optional[int] = Field(default=None)
+    image_pixels_y: Optional[int] = Field(default=None)
+    image_pixel_size: Optional[float] = Field(default=None)
+    thumbnail_pixels_x: Optional[int] = Field(default=None)
+    thumbnail_pixels_y: Optional[int] = Field(default=None)
+    thumbnail_pixel_size: Optional[float] = Field(default=None)
     units: Optional[str] = Field(default=None)
 
     # Extent of the imaged area in real space

src/murfey/util/models.py

@@ -127,22 +127,34 @@ class Base(BaseModel):
 
 class GridSquareParameters(BaseModel):
     tag: str
+    image: str = ""
+
     x_location: Optional[float] = None
-    x_location_scaled: Optional[int] = None
     y_location: Optional[float] = None
+
+    # Image coordinates when overlaid on atlas (in pixels0)
+    x_location_scaled: Optional[int] = None
     y_location_scaled: Optional[int] = None
+
     x_stage_position: Optional[float] = None
     y_stage_position: Optional[float] = None
+
+    # Size of original image (in pixels)
     readout_area_x: Optional[int] = None
     readout_area_y: Optional[int] = None
+
+    # Size of thumbnail used (in pixels)
     thumbnail_size_x: Optional[int] = None
     thumbnail_size_y: Optional[int] = None
+
     height: Optional[int] = None
-    height_scaled: Optional[int] = None
     width: Optional[int] = None
+
+    # Size of image when overlaid on atlas (in pixels)
+    height_scaled: Optional[int] = None
     width_scaled: Optional[int] = None
+
     pixel_size: Optional[float] = None
-    image: str = ""
     angle: Optional[float] = None
 
 

src/murfey/workflows/clem/register_align_and_merge_results.py

@@ -22,7 +22,9 @@ class AlignAndMergeResult(BaseModel):
     align_self: Optional[str] = None
     flatten: Optional[str] = "mean"
     align_across: Optional[str] = None
-    composite_image: Path
+    output_file: Path
+    thumbnail: Optional[Path] = None
+    thumbnail_size: Optional[tuple[int, int]] = None
 
     @field_validator("image_stacks", mode="before")
     @classmethod

src/murfey/workflows/clem/register_preprocessing_results.py

@@ -38,6 +38,10 @@ class CLEMPreprocessingResult(BaseModel):
     output_files: dict[
         Literal["gray", "red", "green", "blue", "cyan", "magenta", "yellow"], Path
     ]
+    thumbnails: dict[
+        Literal["gray", "red", "green", "blue", "cyan", "magenta", "yellow"], Path
+    ] = {}
+    thumbnail_size: Optional[tuple[int, int]] = None  # height, width
     metadata: Path
     parent_lif: Optional[Path] = None
     parent_tiffs: dict[
@@ -54,7 +58,10 @@ class CLEMPreprocessingResult(BaseModel):
 def _is_clem_atlas(result: CLEMPreprocessingResult):
     # If an image has a width/height of at least 1.5 mm, it should qualify as an atlas
     return (
-        max(result.pixels_x * result.pixel_size, result.pixels_y * result.pixel_size)
+        max(
+            result.pixels_x * result.pixel_size,
+            result.pixels_y * result.pixel_size,
+        )
         >= processing_params.atlas_threshold
     )
 
@@ -149,17 +156,29 @@ def _register_clem_image_series(
             murfey_db.commit()
 
     # Add metadata for this series
-    clem_img_series.search_string = str(output_file.parent / "*tiff")
+    clem_img_series.image_search_string = str(output_file.parent / "*tiff")
     clem_img_series.data_type = "atlas" if _is_clem_atlas(result) else "grid_square"
     clem_img_series.number_of_members = result.number_of_members
-    clem_img_series.pixels_x = result.pixels_x
-    clem_img_series.pixels_y = result.pixels_y
-    clem_img_series.pixel_size = result.pixel_size
+    clem_img_series.image_pixels_x = result.pixels_x
+    clem_img_series.image_pixels_y = result.pixels_y
+    clem_img_series.image_pixel_size = result.pixel_size
     clem_img_series.units = result.units
     clem_img_series.x0 = result.extent[0]
     clem_img_series.x1 = result.extent[1]
     clem_img_series.y0 = result.extent[2]
     clem_img_series.y1 = result.extent[3]
+    # Register thumbnails if they are present
+    if result.thumbnails and result.thumbnail_size:
+        thumbnail = list(result.thumbnails.values())[0]
+        clem_img_series.thumbnail_search_string = str(thumbnail.parent / "*.png")
+
+        thumbnail_height, thumbnail_width = result.thumbnail_size
+        scaling_factor = min(
+            thumbnail_height / result.pixels_y, thumbnail_width / result.pixels_x
+        )
+        clem_img_series.thumbnail_pixel_size = result.pixel_size / scaling_factor
+        clem_img_series.thumbnail_pixels_x = int(result.pixels_x * scaling_factor)
+        clem_img_series.thumbnail_pixels_y = int(result.pixels_y * scaling_factor)
     murfey_db.add(clem_img_series)
     murfey_db.commit()
     murfey_db.close()
@@ -189,8 +208,23 @@ def _register_dcg_and_atlas(
     # Determine values for atlas
     if _is_clem_atlas(result):
         output_file = list(result.output_files.values())[0]
-        atlas_name = str(output_file.parent / "*.tiff")
-        atlas_pixel_size = result.pixel_size
+        # Register the thumbnail entries if they are provided
+        if result.thumbnails and result.thumbnail_size is not None:
+            # Glob path to the thumbnail files
+            thumbnail = list(result.thumbnails.values())[0]
+            atlas_name = str(thumbnail.parent / "*.png")
+
+            # Work out the scaling factor used
+            thumbnail_height, thumbnail_width = result.thumbnail_size
+            scaling_factor = min(
+                thumbnail_width / result.pixels_x,
+                thumbnail_height / result.pixels_y,
+            )
+            atlas_pixel_size = result.pixel_size / scaling_factor
+        # Otherwise, register the TIFF files themselves
+        else:
+            atlas_name = str(output_file.parent / "*.tiff")
+            atlas_pixel_size = result.pixel_size
     else:
         atlas_name = ""
         atlas_pixel_size = 0.0
@@ -308,8 +342,6 @@ def _register_grid_square(
             and atlas_entry.x1 is not None
             and atlas_entry.y0 is not None
             and atlas_entry.y1 is not None
-            and atlas_entry.pixels_x is not None
-            and atlas_entry.pixels_y is not None
         ):
             atlas_width_real = atlas_entry.x1 - atlas_entry.x0
             atlas_height_real = atlas_entry.y1 - atlas_entry.y0
@@ -318,32 +350,40 @@ def _register_grid_square(
             return
 
         for clem_img_series in clem_img_series_to_register:
+            # Register datasets using thumbnail sizes and scales
             if (
                 clem_img_series.x0 is not None
                 and clem_img_series.x1 is not None
                 and clem_img_series.y0 is not None
                 and clem_img_series.y1 is not None
+                and clem_img_series.thumbnail_pixels_x is not None
+                and clem_img_series.thumbnail_pixels_y is not None
+                and clem_img_series.thumbnail_pixel_size is not None
             ):
                 # Find pixel corresponding to image midpoint on atlas
                 x_mid_real = (
                     0.5 * (clem_img_series.x0 + clem_img_series.x1) - atlas_entry.x0
                 )
-                x_mid_px = int(x_mid_real / atlas_width_real * atlas_entry.pixels_x)
+                x_mid_px = int(
+                    x_mid_real / atlas_width_real * clem_img_series.thumbnail_pixels_x
+                )
                 y_mid_real = (
                     0.5 * (clem_img_series.y0 + clem_img_series.y1) - atlas_entry.y0
                 )
-                y_mid_px = int(y_mid_real / atlas_height_real * atlas_entry.pixels_y)
+                y_mid_px = int(
+                    y_mid_real / atlas_height_real * clem_img_series.thumbnail_pixels_y
+                )
 
-                # Find the number of pixels in width and height the image corresponds to on the atlas
+                # Find the size of the image, in pixels, when overlaid the atlas
                 width_scaled = int(
                     (clem_img_series.x1 - clem_img_series.x0)
                     / atlas_width_real
-                    * atlas_entry.pixels_x
+                    * clem_img_series.thumbnail_pixels_x
                 )
                 height_scaled = int(
                     (clem_img_series.y1 - clem_img_series.y0)
                     / atlas_height_real
-                    * atlas_entry.pixels_y
+                    * clem_img_series.thumbnail_pixels_y
                 )
             else:
                 logger.warning(
@@ -358,14 +398,18 @@ def _register_grid_square(
                 x_location_scaled=x_mid_px,
                 y_location=clem_img_series.y0,
                 y_location_scaled=y_mid_px,
-                height=clem_img_series.pixels_x,
-                height_scaled=height_scaled,
-                width=clem_img_series.pixels_y,
+                readout_area_x=clem_img_series.image_pixels_x,
+                readout_area_y=clem_img_series.image_pixels_y,
+                thumbnail_size_x=clem_img_series.thumbnail_pixels_x,
+                thumbnail_size_y=clem_img_series.thumbnail_pixels_y,
+                width=clem_img_series.image_pixels_x,
                 width_scaled=width_scaled,
-                x_stage_position=clem_img_series.x0,
-                y_stage_position=clem_img_series.y0,
-                pixel_size=clem_img_series.pixel_size,
-                image=clem_img_series.search_string,
+                height=clem_img_series.image_pixels_y,
+                height_scaled=height_scaled,
+                x_stage_position=0.5 * (clem_img_series.x0 + clem_img_series.x1),
+                y_stage_position=0.5 * (clem_img_series.y0 + clem_img_series.y1),
+                pixel_size=clem_img_series.image_pixel_size,
+                image=clem_img_series.thumbnail_search_string,
             )
             # Register or update the grid square entry as required
             if grid_square_result := murfey_db.exec(

tests/workflows/clem/test_register_preprocessing_results.py

@@ -77,6 +77,14 @@ def generate_preprocessing_messages(
         output_files = {color: str(series_path / f"{color}.tiff") for color in colors}
         for output_file in output_files.values():
             Path(output_file).touch(exist_ok=True)
+        thumbnails = {
+            color: str(series_path / ".thumbnails" / f"{color}.png") for color in colors
+        }
+        for v in thumbnails.values():
+            if not (thumbnail := Path(v)).parent.exists():
+                thumbnail.parent.mkdir(parents=True)
+            thumbnail.touch(exist_ok=True)
+        thumbnail_size = (512, 512)
         is_stack = dataset[1]
         is_montage = dataset[2]
         shape = dataset[3]
@@ -91,6 +99,8 @@ def generate_preprocessing_messages(
                 "is_stack": is_stack,
                 "is_montage": is_montage,
                 "output_files": output_files,
+                "thumbnails": thumbnails,
+                "thumbnail_size": thumbnail_size,
                 "metadata": str(metadata),
                 "parent_lif": None,
                 "parent_tiffs": {},