feat(tasks): Refactor series_reconstruct_task and add _stamp_derived_dataset for improved dataset handling

atlas/tasks.py

@@ -86,7 +86,6 @@ from generic.models import CimarCase
 from rad.settings import REMOTE_URL, CIMAR_USERNAME, CIMAR_PASSWORD
 from helpers.cimar import CimarAPI, NotFoundError
 from pydicom.uid import generate_uid
-from datetime import datetime
 from django.contrib.auth.models import User
 from django.core.files.base import ContentFile
 from django.core.cache import cache
@@ -392,9 +391,6 @@ def series_reconstruct_task(
         recon_series = atlas_views._create_series_derivative(series, f"Recon {plane_norm.title()} ({source_plane.title()} src)")
         recon_series.series_instance_uid = generate_uid()
         recon_series.save(update_fields=["series_instance_uid"])
-        now = datetime.utcnow()
-        date_str = now.strftime("%Y%m%d")
-        time_str = now.strftime("%H%M%S.%f")[:13]
 
         for idx in range(int(cfg["count"])):
             arr2d = cfg["slice_fn"](idx)
@@ -404,18 +400,16 @@ def series_reconstruct_task(
             ds_new.Rows = int(arr2d.shape[0])
             ds_new.Columns = int(arr2d.shape[1])
             ds_new.InstanceNumber = idx + 1
-            ds_new.SOPInstanceUID = generate_uid()
+            atlas_views._stamp_derived_dataset(
-            ds_new.SeriesInstanceUID = recon_series.series_instance_uid
+                ds_new,
-            ds_new.SeriesDescription = recon_series.description or f"Recon {plane_norm.title()}"
+                series_instance_uid=recon_series.series_instance_uid,
-            ds_new.ImageType = ["DERIVED", "SECONDARY", "MPR"]
+                series_description=recon_series.description or f"Recon {plane_norm.title()}",
-            ds_new.DerivationDescription = (
+                derivation_description=(
                 f"{plane_norm.title()} reconstruction; mode={recon_thickness_mode}; "
                 f"thickness={float(cfg['slice_thickness_out']):.3f}mm; spacing={float(cfg['spacing_between_slices_out']):.3f}mm"
+                ),
+                image_type=["DERIVED", "SECONDARY", "MPR"],
             )
-            ds_new.ContentDate = date_str
-            ds_new.ContentTime = time_str
-            ds_new.InstanceCreationDate = date_str
-            ds_new.InstanceCreationTime = time_str
             ds_new.PixelData = arr2d.tobytes()
             ds_new.PixelSpacing = [float(cfg["pixel_spacing_out"][0]), float(cfg["pixel_spacing_out"][1])]
             ds_new.ImageOrientationPatient = cfg["image_orientation_out"]

atlas/templates/atlas/series_viewer.html

@@ -212,7 +212,7 @@
                             <button class="btn btn-outline-warning btn-sm"
                                     data-bs-toggle="modal"
                                     data-bs-target="#truncate-series-modal">
-                                <i class="bi bi-scissors"></i> Truncate series
+                                <i class="bi bi-sliders"></i> Series tools
                             </button>
 
                             <div class="card mt-2">
@@ -255,7 +255,7 @@
             <div class="modal-dialog modal-dialog-centered modal-xl">
                 <div class="modal-content">
                     <div class="modal-header">
-                        <h5 class="modal-title">Series Optimization</h5>
+                        <h5 class="modal-title">Series Tools</h5>
                         <button type="button" class="btn-close" data-bs-dismiss="modal" aria-label="Close"></button>
                     </div>
                     <div class="modal-body">

atlas/views.py

@@ -307,7 +307,45 @@ def _resize_array_nearest(arr, target_rows, target_cols):
     return arr[np.ix_(row_idx, col_idx)]
 
 
-def _downsample_dicom_dataset(ds, reduction_pct):
+def _stamp_derived_dataset(
+    ds,
+    *,
+    series_instance_uid=None,
+    series_description=None,
+    derivation_description=None,
+    image_type=None,
+    regenerate_sop_instance_uid=True,
+    stamp_series_datetime=True,
+):
+    now = timezone.now()
+    date_str = now.strftime("%Y%m%d")
+    time_str = now.strftime("%H%M%S.%f")[:13]
+
+    if regenerate_sop_instance_uid:
+        ds.SOPInstanceUID = generate_uid()
+    if series_instance_uid:
+        ds.SeriesInstanceUID = series_instance_uid
+    if series_description:
+        ds.SeriesDescription = series_description
+
+    ds.ImageType = image_type or ["DERIVED", "SECONDARY"]
+
+    if derivation_description:
+        ds.DerivationDescription = derivation_description
+
+    ds.ContentDate = date_str
+    ds.ContentTime = time_str
+    ds.InstanceCreationDate = date_str
+    ds.InstanceCreationTime = time_str
+
+    if stamp_series_datetime:
+        ds.SeriesDate = date_str
+        ds.SeriesTime = time_str
+
+    return ds
+
+
+def _downsample_dicom_dataset(ds, reduction_pct, *, series_instance_uid=None, series_description=None):
     if reduction_pct <= 0 or reduction_pct >= 100:
         raise ValueError("Downsample percentage must be between 1 and 99")
 
@@ -321,11 +359,36 @@ def _downsample_dicom_dataset(ds, reduction_pct):
     target_cols = max(1, int(round(px.shape[1] * factor)))
     resized = _resize_array_nearest(px, target_rows, target_cols)
 
+    derivation_description = f"Downsampled by {reduction_pct}% using nearest-neighbour resize"
+    _stamp_derived_dataset(
+        ds,
+        series_instance_uid=series_instance_uid,
+        series_description=series_description,
+        derivation_description=derivation_description,
+        image_type=["DERIVED", "SECONDARY"],
+    )
+
+    if hasattr(ds, "PixelSpacing") and ds.PixelSpacing and len(ds.PixelSpacing) >= 2:
+        row_spacing = _safe_float(ds.PixelSpacing[0], 1.0)
+        col_spacing = _safe_float(ds.PixelSpacing[1], 1.0)
+        ds.PixelSpacing = [float(row_spacing / factor), float(col_spacing / factor)]
+
+    if hasattr(ds, "ImagerPixelSpacing") and ds.ImagerPixelSpacing and len(ds.ImagerPixelSpacing) >= 2:
+        row_spacing = _safe_float(ds.ImagerPixelSpacing[0], 1.0)
+        col_spacing = _safe_float(ds.ImagerPixelSpacing[1], 1.0)
+        ds.ImagerPixelSpacing = [float(row_spacing / factor), float(col_spacing / factor)]
+
+    if hasattr(ds, "NominalScannedPixelSpacing") and ds.NominalScannedPixelSpacing and len(ds.NominalScannedPixelSpacing) >= 2:
+        row_spacing = _safe_float(ds.NominalScannedPixelSpacing[0], 1.0)
+        col_spacing = _safe_float(ds.NominalScannedPixelSpacing[1], 1.0)
+        ds.NominalScannedPixelSpacing = [float(row_spacing / factor), float(col_spacing / factor)]
+
     ds.Rows = int(target_rows)
     ds.Columns = int(target_cols)
     ds.PixelData = resized.tobytes()
     ds.pop("SmallestImagePixelValue", None)
     ds.pop("LargestImagePixelValue", None)
+    ds.pop("PixelAspectRatio", None)
     return ds
 
 
@@ -606,7 +669,12 @@ def series_optimize_htmx(request, series_id):
                     continue
 
                 try:
-                    ds_out = _downsample_dicom_dataset(ds, downsample_pct)
+                    ds_out = _downsample_dicom_dataset(
+                        ds,
+                        downsample_pct,
+                        series_instance_uid=preview_series.series_instance_uid,
+                        series_description=preview_series.description,
+                    )
                 except Exception as exc:
                     logger.warning("Downsample preview failed for image {}: {}", image.pk, exc)
                     continue
@@ -645,6 +713,7 @@ def series_optimize_htmx(request, series_id):
             )
 
         downsampled = []
+        replacement_series_uid = generate_uid()
         for image in full_series_images:
             ds = _read_series_image_dataset(image)
             if ds is None:
@@ -652,7 +721,12 @@ def series_optimize_htmx(request, series_id):
 
             try:
                 original_hash = image.image_blake3_hash
-                ds_out = _downsample_dicom_dataset(ds, downsample_pct)
+                ds_out = _downsample_dicom_dataset(
+                    ds,
+                    downsample_pct,
+                    series_instance_uid=replacement_series_uid,
+                    series_description=series.description,
+                )
                 out_io = io.BytesIO()
                 ds_out.save_as(out_io, write_like_original=False)
                 out_io.seek(0)
@@ -671,7 +745,8 @@ def series_optimize_htmx(request, series_id):
 
         if downsampled:
             series.modified = Series.SeriesModifiedChocies.RE
-            series.save(update_fields=["modified"])
+            series.series_instance_uid = replacement_series_uid
+            series.save(update_fields=["modified", "series_instance_uid"])
 
         return HttpResponse(
             f'<div class="alert alert-success mb-0">Downsample complete ({downsample_pct}%). Updated <strong>{len(downsampled)}</strong> images while preserving original hashes.</div>'