imagedata 3.7.2.dev1__tar.gz → 3.8.0__tar.gz

{imagedata-3.7.2.dev1 → imagedata-3.8.0}/CHANGELOG.md

@@ -7,7 +7,82 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 <!--next-version-placeholder-->
 
-## [v3.7.2-dev1] - 2024-10-23
+## [v3.8.0] - 2025-04-28
+### Added
+* Novel sorting routine for n-dimensional DICOM datasets.
+* Allow user-defined sorting criteria, and overriding default sorting.
+* Sort on Trigger Time also.
+* Added code to sort series on diffusion gradient direction (b-vector),
+  and specifically diffusion RSI data.
+
+### Changed
+* Image tags property (tags) is now a multi-dimensional array when the image dimension > 4.
+  In this case the image tags will be tuples.
+* diffusion.py: Added code to get diffusion b value from Siemens E11 format.
+* Patient class: Limit the strict check of patient attributes to
+  patientName, patientID, patientBirthDate
+  and patientSex.
+* Series.__getitem__(): Accept tuple slicing specification, like ((2,),(2,),(2,)).
+* Improved some test suites to run better on Windows.
+
+### Fixed
+* Corrected parsing windows UNC path names with leading double slash.
+* ZipFileArchive: Windows: Do not complain when temporary local file cannot be removed.
+* XnatTransport.open(): Corrected scan search to search for series description,
+  not series number.
+* Viewer.viewport_set(): Fixed error where the viewport did not include the last image.
+* evidence2roi: Do not raise exception for unknown roi type. Log a warning instead.
+
+## [v3.8.0-rc3] - 2025-04-28
+* Added sort on TriggerTime
+
+## [v3.8.0-rc2] - 2025-04-23
+* Allow user-defined sorting criteria, and overriding default sorting.
+
+## [v3.8.0-rc1] - 2025-04-11
+* Novel sorting routine for n-dimensional DICOM datasets.
+
+## [v3.7.3-dev8] - 2025-03-07
+* Corrected parsing windows UNC path names with leading double slash.
+
+## [v3.7.3-dev7] - 2025-03-07
+* ZipFileArchive: Windows: Do not complain when temporary local file cannot be removed.
+* Improved some test suites to run better on Windows.
+
+## [v3.7.3-rc0] - 2025-01-06
+* Release candidate 3.7.3-rc0
+
+## [v3.7.3-dev6] - 2024-12-13
+* Series.__getitem__(): Accept tuple slicing specification, like ((2,),(2,),(2,)).
+
+## [v3.7.3-dev5] - 2024-12-12
+* Patient class: Limit the strict check of patient attributes to
+  patientName, patientID, patientBirthDate
+  and patientSex.
+
+## [v3.7.3-dev4] - 2024-12-04
+* evidence2roi: Do not raise exception for unknown roi type. Log a warning instead.
+
+## [v3.7.3-dev3] - 2024-11-26
+* Working on n-dimensional sorting.
+* Added code to sort series on diffusion gradient direction (b-vector),
+  and specifically diffusion RSI data.
+
+## [v3.7.3-dev2] - 2024-11-25
+* diffusion.py: Added code to get diffusion b value from Siemens E11 format.
+
+## [v3.7.3-dev1] - 2024-11-06
+* Viewer.viewport_set(): Fixed error where the viewport did not include the last image.
+
+## [v3.7.3-dev0] - 2024-11-04
+* XnatTransport.open(): Corrected scan search to search for series description,
+  not series number.
+
+## [v3.7.2] - 2024-10-28
+* Fixed viewing in jupyter notebook using backend `widget`.
+* Renamed collections.py to collection.py.
+* 
+## [v3.7.2-dev2] - 2024-10-25
 * Fixed viewing in jupyter notebook using backend `widget`.
 
 ## [v3.7.2-dev0] - 2024-10-10

{imagedata-3.7.2.dev1 → imagedata-3.8.0}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.4
 Name: imagedata
-Version: 3.7.2.dev1
+Version: 3.8.0
 Summary: Read/write medical image data
 Author-email: Erling Andersen <Erling.Andersen@Helse-Bergen.NO>
 Maintainer-email: Erling Andersen <Erling.Andersen@Helse-Bergen.NO>
@@ -63,6 +63,7 @@ Requires-Dist: pillow>=10.0.0
 Requires-Dist: xnat
 Requires-Dist: matplotlib>=3.8.0
 Requires-Dist: opencv-python-headless
+Dynamic: license-file
 
 #########
 imagedata

imagedata-3.8.0/VERSION.txt

@@ -0,0 +1 @@
+3.8.0

{imagedata-3.7.2.dev1 → imagedata-3.8.0}/docs/Options.rst

@@ -61,6 +61,11 @@ The following plugin options are known at the time of writing:
 |                         |                         |     |times for dynamic      |
 |                         |                         |     |series                 |
 +-------------------------+-------------------------+-----+-----------------------+
+|read                     |ignore_series_uid        |str  |Ignore Series Instance |
+|                         |                         |     |UID, i.e. do not sort  |
+|                         |                         |     |images into different  |
+|                         |                         |     |Series                 |
++-------------------------+-------------------------+-----+-----------------------+
 |read                     |input_serinsuid          |str  |Filter input files on  |
 |                         |                         |     |specified              |
 |                         |                         |     |Series Instance UID    |

imagedata-3.8.0/docs/Sorting.rst

@@ -0,0 +1,129 @@
+.. _Sorting:
+
+Sorting
+=======
+
+When reading data files to construct a Series instance, the images will be sorted into a
+multidimensional array.
+In particular, DICOM series typically comes as 2D slice images.
+These images are sorted into 3D volumes on the basis of their geometrical information
+(image position, or slice location).
+
+The following example loads a 3D volume from 'dicom/volume' and sorts the images into
+a 3D volume of 40 slices.
+
+.. code-block:: python
+
+    img = Series('dicom/volume')
+    print(img.shape)
+    >>> (40, 192, 152)
+
+4D data can be read by sorting the volumes on some DICOM attribute. In the following example, a
+4D dataset is loaded and volumes are sorted according to Acquisition Time. Notice that there
+are 10 time steps and 40 slices in the dataset.
+
+.. code-block:: python
+
+    img = Series('dicom/time', input_order='time')
+    print(img.shape)
+    >>> (10, 40, 192, 152)
+
+Several sort criteria are predefined:
+
+* none: No sorting (2D/3D datasets only)
+* auto: Determine sorting criteria automatically (default) (see below)
+* time: Sort on Acquisition Time
+* triggertime: Sort on Trigger Time
+* b: Sort on MR diffusion b-value
+* bvector: Sort on MR diffusion b vector
+* fa: Sort on MR Flip Angle
+* te: Sort on MR Echo Time
+
+Auto-sorting
+------------
+
+`Imagedata` can determine the sorting automatically in some typical cases,
+attempting to sort on `time`, `b`-value, echo time or flip angle.
+
+The exact list of sorting criteria can be set using the `auto_sort` list.
+The default auto sort list is ['time', 'triggertime', 'b', 'fa', 'te'].
+
+.. code-block:: python
+
+    img = Series('dicomdata', auto_sort=['time', 'b'])
+
+User-defined sorting criteria
+-----------------------------
+
+When your data does not match any of these sort criteria, a new sort criteria can be defined.
+E.g., sorting MR images on Inversion Time ('ti'), a new sort criteria 'ti' is coupled to the
+DICOM attribute Inversion Time:
+
+.. code-block:: python
+
+    img = Series('dicom/time', input_order='ti', ti='InversionTime')
+    print(img.shape)
+    >>> (10, 40, 192, 152)
+
+More advanced extraction of sorting values can be implemented creating a
+user function which returns a value for each Dataset:
+
+.. code-block:: python
+
+    def get_TI(im: Dataset) -> float:
+        return float(im.data_element('InversionTime').value)
+
+    img = Series('dicom/TIdata', input_order='ti', ti=get_TI)
+
+A similar user function to calculate time in seconds from Trigger Time (ms).
+Notice how the get_TriggerTime() function overloads the standard `time`
+definition. This allow to treat the acquired series as a time series with
+the `timeline` property.
+
+NOTICE: Trigger Time is now implemented in the standard library.
+This discussion remains here to document the possible uses.
+
+.. code-block:: python
+
+    def get_TriggerTime(im: Dataset) -> float:
+        return float(im.data_element('TriggerTime').value / 1000.)
+
+    img = Series('dicom/triggerTime', input_order='time', time=get_TriggerTime)
+    img.timeline
+
+Auto-sorting either on Acquisition Time or Trigger Time can be implemented.
+In this case, the resulting series will not have the `timeline` property when
+Trigger Time is the sorting criteria:
+
+.. code-block:: python
+
+    img = Series('dicomdata', auto_sort=['time', 'trigger'],
+        input_order='trigger', trigger=get_TriggerTime)
+    )
+
+N-dimensional sorting
+---------------------
+
+While 4D data can be sorted automatically, higher dimensions must be defined explicitly.
+The `input_order` parameter can be a comma-separated list of sorting criteria.
+
+A dynamic dual-echo MR acquisition can be sorted on time and echo time into a 5D Series object, like:
+
+.. code-block:: python
+
+    img = Series('dyn_dual_echo', input_order='time,te')
+
+In particular, MR RSI diffusion data can be sorted on `b` value and `b` vector:
+
+.. code-block:: python
+
+    img = Series('diff_rsi', input_order='b,bvector')
+    tags = img.tags[0]
+    for idx in np.ndindex(tags.shape):
+        try:
+            b, bvector = tags[idx]
+        except TypeError:
+            continue
+        rsi = img[idx]
+        print(b, bvector, rsi.shape)
+

{imagedata-3.7.2.dev1 → imagedata-3.8.0}/src/imagedata/apps/Siemens/evidence2mask.py

@@ -135,8 +135,28 @@ def evidence2roi(im, uid_table=None, content=None):
             roi_name = finding[(0x0029, 0x1030)].value
         except KeyError:
             roi_name = 'NONAME'
+        if roi_type_value not in (
+            'PolygonApplication3D', 'FreehandApplication3D',
+            'MarkerApplication3D', 'EllipseApplication3D',
+            'StandaloneTextApplication3D', 'DistanceLineApplication3D'
+        ):
+            # Attempt to find roi type in advanced presentation sequence (0029,1091)
+            try:
+                adv_pres_seq = finding[(0x0029, 0x1091)][0]
+                try:
+                    roi_type_value = adv_pres_seq[(0x0029, 0x108e)].value.decode()
+                    roi_name = finding[(0x0029, 0x1032)].value.decode()
+                except AttributeError:
+                    roi_type_value = adv_pres_seq[(0x0029, 0x108e)].value
+                    roi_name = finding[(0x0029, 0x1032)].value
+            except KeyError:
+                pass
         logger.info("Finding: {} {} {}".format(content['creator'], roi_name, roi_type_value))
-        if roi_type_value == 'PolygonApplication3D' or roi_type_value == 'FreehandApplication3D':
+        if roi_type_value in ('PolygonApplication3D', 'FreehandApplication3D', 'ROI2D'):
+            polygon = _get_measurement_points(finding[(0x0029, 0x1096)], zyx=True)
+            rois.append(PolygonROI(polygon, roi_name, stu_ins_uid, ser_ins_uid, sop_ins_uid))
+            logger.debug('ROI {}: {} points'.format(meas_appl_number, len(polygon) // 3))
+        elif roi_type_value == 'ROI2D':
             polygon = _get_measurement_points(finding[(0x0029, 0x1096)], zyx=True)
             rois.append(PolygonROI(polygon, roi_name, stu_ins_uid, ser_ins_uid, sop_ins_uid))
             logger.debug('ROI {}: {} points'.format(meas_appl_number, len(polygon) // 3))
@@ -159,7 +179,9 @@ def evidence2roi(im, uid_table=None, content=None):
             logger.warning("DistanceLineApplication3D ROI not implemented.")
             pass
         else:
-            raise ValueError("ROI type {} ({}) not implemented.".format(
-                roi_type_value, type(roi_type_value)))
+            logging.warning(
+                "ROI type {} ({}) not implemented.".format(
+                    roi_type_value, type(roi_type_value))
+            )
 
     return rois, content