nettracer3d 0.6.2__tar.gz → 0.6.3__tar.gz

{nettracer3d-0.6.2/src/nettracer3d.egg-info → nettracer3d-0.6.3}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.4
 Name: nettracer3d
-Version: 0.6.2
+Version: 0.6.3
 Summary: Scripts for intializing and analyzing networks from segmentations of three dimensional images.
 Author-email: Liam McLaughlin <mclaughlinliam99@gmail.com>
 Project-URL: User_Tutorial, https://www.youtube.com/watch?v=cRatn5VTWDY
@@ -33,6 +33,7 @@ Provides-Extra: cuda12
 Requires-Dist: cupy-cuda12x; extra == "cuda12"
 Provides-Extra: cupy
 Requires-Dist: cupy; extra == "cupy"
+Dynamic: license-file
 
 NetTracer3D is a python package developed for both 2D and 3D analysis of microscopic images in the .tif file format. It supports generation of 3D networks showing the relationships between objects (or nodes) in three dimensional space, either based on their own proximity or connectivity via connecting objects such as nerves or blood vessels. In addition to these functionalities are several advanced 3D data processing algorithms, such as labeling of branched structures or abstraction of branched structures into networks. Note that nettracer3d uses segmented data, which can be segmented from other softwares such as ImageJ and imported into NetTracer3D, although it does offer its own segmentation via intensity and volumetric thresholding, or random forest machine learning segmentation. NetTracer3D currently has a fully functional GUI. To use the GUI, after installing the nettracer3d package via pip, enter the command 'nettracer3d' in your command prompt:
 
@@ -44,8 +45,10 @@ NetTracer3D is free to use/fork for academic/nonprofit use so long as citation i
 
 NetTracer3D was developed by Liam McLaughlin while working under Dr. Sanjay Jain at Washington University School of Medicine.
 
--- Version 0.6.2 updates --
+-- Version 0.6.3 updates --
 
-1. Fixed bug with performing 2D distance transforms on CPU
+1. Fixed bug with the active channel indicator on the GUI not always updating when the active channel was changed by internal loading.
 
-2. Updated ram_lock mode in the segmenter to use smaller chunks and garbage collect better.
+2. Updated ram_lock mode in the segmenter to garbage collect better... again.
+
+3. Added new function (Label neighborhoods). Find it in process -> image -> label neighborhoods. Allows a 3d labeled array to act as a kernel and extend its labels along a secondary binary array. This can be useful in evaluating nearest neighbors en-masse. The method behind this (smart label) was previously used internally for some methods (ie watershedding) but I thought I'd include it as an actual function since it has some more general uses.

{nettracer3d-0.6.2 → nettracer3d-0.6.3}/README.md

@@ -8,8 +8,10 @@ NetTracer3D is free to use/fork for academic/nonprofit use so long as citation i
 
 NetTracer3D was developed by Liam McLaughlin while working under Dr. Sanjay Jain at Washington University School of Medicine.
 
--- Version 0.6.2 updates --
+-- Version 0.6.3 updates --
 
-1. Fixed bug with performing 2D distance transforms on CPU
+1. Fixed bug with the active channel indicator on the GUI not always updating when the active channel was changed by internal loading.
 
-2. Updated ram_lock mode in the segmenter to use smaller chunks and garbage collect better.
+2. Updated ram_lock mode in the segmenter to garbage collect better... again.
+
+3. Added new function (Label neighborhoods). Find it in process -> image -> label neighborhoods. Allows a 3d labeled array to act as a kernel and extend its labels along a secondary binary array. This can be useful in evaluating nearest neighbors en-masse. The method behind this (smart label) was previously used internally for some methods (ie watershedding) but I thought I'd include it as an actual function since it has some more general uses.

{nettracer3d-0.6.2 → nettracer3d-0.6.3}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "nettracer3d"
-version = "0.6.2"
+version = "0.6.3"
 authors = [
   { name="Liam McLaughlin", email="mclaughlinliam99@gmail.com" },
 ]

{nettracer3d-0.6.2 → nettracer3d-0.6.3}/src/nettracer3d/nettracer_gui.py

@@ -2429,10 +2429,14 @@ class ImageViewerWindow(QMainWindow):
         binarize_action.triggered.connect(self.show_binarize_dialog)
         label_action = image_menu.addAction("Label Objects")
         label_action.triggered.connect(self.show_label_dialog)
+        slabel_action = image_menu.addAction("Neighborhood Labels")
+        slabel_action.triggered.connect(self.show_slabel_dialog)
         thresh_action = image_menu.addAction("Threshold/Segment")
         thresh_action.triggered.connect(self.show_thresh_dialog)
         mask_action = image_menu.addAction("Mask Channel")
         mask_action.triggered.connect(self.show_mask_dialog)
+        type_action = image_menu.addAction("Channel dtype")
+        type_action.triggered.connect(self.show_type_dialog)
         skeletonize_action = image_menu.addAction("Skeletonize")
         skeletonize_action.triggered.connect(self.show_skeletonize_dialog)
         watershed_action = image_menu.addAction("Watershed")
@@ -2657,6 +2661,11 @@ class ImageViewerWindow(QMainWindow):
         dialog = LabelDialog(self)
         dialog.exec()
 
+    def show_slabel_dialog(self):
+        """Show the slabel dialog"""
+        dialog = SLabelDialog(self)
+        dialog.exec()
+
     def show_thresh_dialog(self):
         """Show threshold dialog"""
         if self.machine_window is not None:
@@ -2671,6 +2680,11 @@ class ImageViewerWindow(QMainWindow):
         dialog = MaskDialog(self)
         dialog.exec()
 
+    def show_type_dialog(self):
+        """Show the type dialog"""
+        dialog = TypeDialog(self)
+        dialog.exec()
+
     def show_skeletonize_dialog(self):
         """show the skeletonize dialog"""
         dialog = SkeletonizeDialog(self)
@@ -3066,6 +3080,7 @@ class ImageViewerWindow(QMainWindow):
     def set_active_channel(self, index):
         """Set the active channel and update UI accordingly."""
         self.active_channel = index
+        self.active_channel_combo.setCurrentIndex(index)
         # Update button appearances to show active channel
         for i, btn in enumerate(self.channel_buttons):
             if i == index and btn.isEnabled():
@@ -6261,6 +6276,91 @@ class LabelDialog(QDialog):
                 f"Error running label: {str(e)}"
             )
 
+
+class SLabelDialog(QDialog):
+    def __init__(self, parent=None):
+        super().__init__(parent)
+        self.setWindowTitle("Smart Label (Use label array to assign label neighborhoods to binary array)?")
+        self.setModal(True)
+        
+        layout = QFormLayout(self)
+
+
+        # Add mode selection dropdown
+        self.mode_selector = QComboBox()
+        self.mode_selector.addItems(["Nodes", "Edges", "Overlay 1", "Overlay 2"])
+        self.mode_selector.setCurrentIndex(0)  # Default to Mode 1
+        layout.addRow("Prelabeled Array:", self.mode_selector)
+
+        layout.addRow(QLabel("Will Label Neighborhoods in: "))
+
+        # Add mode selection dropdown
+        self.target_selector = QComboBox()
+        self.target_selector.addItems(["Nodes", "Edges", "Overlay 1", "Overlay 2"])
+        self.target_selector.setCurrentIndex(1)  # Default to Mode 1
+        layout.addRow("Binary Array:", self.target_selector)
+
+        # GPU checkbox (default True)
+        self.GPU = QPushButton("GPU")
+        self.GPU.setCheckable(True)
+        self.GPU.setChecked(True)
+        layout.addRow("Use GPU:", self.GPU)
+
+        self.down_factor = QLineEdit("")
+        layout.addRow("Internal Downsample for GPU (if needed):", self.down_factor)
+
+       # Add Run button
+        run_button = QPushButton("Run Smart Label")
+        run_button.clicked.connect(self.run_slabel)
+        layout.addRow(run_button)
+
+    def run_slabel(self):
+
+        try:
+
+            accepted_source = self.mode_selector.currentIndex()
+            accepted_target = self.target_selector.currentIndex()
+            GPU = self.GPU.isChecked()
+
+
+            if accepted_source == accepted_target:
+                return
+
+            binary_array = self.parent().channel_data[accepted_target]
+
+            label_array = self.parent().channel_data[accepted_source]
+
+            down_factor = float(self.down_factor.text()) if self.down_factor.text().strip() else None
+
+
+            try:
+
+                # Update both the display data and the network object
+                binary_array = sdl.smart_label(binary_array, label_array, directory = None, GPU = GPU, predownsample = down_factor)
+
+                label_array = sdl.invert_array(label_array)
+
+                binary_array = binary_array * label_array
+
+                self.parent().load_channel(accepted_target, binary_array, True)
+
+                self.accept()
+                
+            except Exception as e:
+                QMessageBox.critical(
+                    self,
+                    "Error",
+                    f"Error running smart label: {str(e)}"
+                )
+
+        except Exception as e:
+            QMessageBox.critical(
+                self,
+                "Error",
+                f"Error running smart label: {str(e)}"
+            )
+
+
 class ThresholdDialog(QDialog):
     def __init__(self, parent=None):
         super().__init__(parent)
@@ -6380,9 +6480,8 @@ class MachineWindow(QMainWindow):
                 act_channel = 1
 
 
-        array1 = np.zeros_like(active_data)
-        array2 = np.zeros_like(active_data)
-        array3 = np.zeros_like(active_data)
+        array1 = np.zeros_like(active_data).astype(np.uint8)
+        array3 = np.zeros_like(active_data).astype(np.uint8)
         self.parent().highlight_overlay = array3 #Clear this out for the segmenter to use
 
         self.parent().load_channel(2, array1, True)
@@ -6460,7 +6559,7 @@ class MachineWindow(QMainWindow):
 
         # Group 3: Training Options
         training_group = QGroupBox("Training")
-        training_layout = QVBoxLayout()
+        training_layout = QHBoxLayout()
         train_quick = QPushButton("Train Quick Model")
         train_quick.clicked.connect(lambda: self.train_model(speed=True))
         train_detailed = QPushButton("Train More Detailed Model")
@@ -6471,11 +6570,13 @@ class MachineWindow(QMainWindow):
 
         # Group 4: Segmentation Options
         segmentation_group = QGroupBox("Segmentation")
-        segmentation_layout = QVBoxLayout()
+        segmentation_layout = QHBoxLayout()
         seg_button = QPushButton("Preview Segment")
         self.seg_button = seg_button
         seg_button.clicked.connect(self.start_segmentation)
-        self.lock_button = QPushButton("🔒 Memory lock - (Prioritize RAM. Recommended unless you have a lot)")
+        self.pause_button = QPushButton("▶/⏸️")
+        self.pause_button.clicked.connect(self.pause)
+        self.lock_button = QPushButton("🔒 Memory lock - (Prioritize RAM)")
         self.lock_button.setCheckable(True)
         self.lock_button.setChecked(True)
         self.lock_button.clicked.connect(self.toggle_lock)
@@ -6483,6 +6584,7 @@ class MachineWindow(QMainWindow):
         full_button = QPushButton("Segment All")
         full_button.clicked.connect(self.segment)
         segmentation_layout.addWidget(seg_button)
+        #segmentation_layout.addWidget(self.pause_button)   # <--- for some reason the segmenter preview is still running even when killed, may be regenerating itself somewhere. May or may not actually try to resolve this because this feature isnt that necessary.
         segmentation_layout.addWidget(self.lock_button)
         segmentation_layout.addWidget(full_button)
         segmentation_group.setLayout(segmentation_layout)
@@ -6507,6 +6609,27 @@ class MachineWindow(QMainWindow):
 
         self.mem_lock = self.lock_button.isChecked()
 
+    def pause(self):
+
+        if self.segmentation_worker is not None:
+            try:
+                print("Pausing segmenter")
+                self.previewing = False
+                self.segmentation_finished
+                del self.segmentation_worker
+                self.segmentation_worker = None
+            except:
+                pass
+
+        else:
+            try:
+                print("Restarting segmenter")
+                self.previewing = True
+                self.start_segmentation
+            except:
+                pass
+
+
 
     def toggle_GPU(self):
 
@@ -6615,7 +6738,7 @@ class MachineWindow(QMainWindow):
             else:
                 active_data = self.parent().channel_data[1]
 
-        array3 = np.zeros_like(active_data)
+        array3 = np.zeros_like(active_data).astype(np.uint8)
         self.parent().highlight_overlay = array3 #Clear this out for the segmenter to use
 
         if not self.trained:
@@ -6801,7 +6924,7 @@ class MachineWindow(QMainWindow):
                 else:
                     active_data = self.parent().channel_data[1]
 
-            array3 = np.zeros_like(active_data)
+            array3 = np.zeros_like(active_data).astype(np.uint8)
             self.parent().highlight_overlay = array3 #Clear this out for the segmenter to use
 
             print("Segmenting entire volume with model...")
@@ -7589,6 +7712,65 @@ class MaskDialog(QDialog):
             print(f"Error masking: {e}")
 
 
+class TypeDialog(QDialog):
+
+    def __init__(self, parent=None):
+
+        super().__init__(parent)
+        self.setWindowTitle("Active Channel dtype")
+        self.setModal(True)
+
+        layout = QFormLayout(self)
+
+        self.active_chan = self.parent().active_channel
+
+        active_data = self.parent().channel_data[self.active_chan]
+
+        layout.addRow("Info:", QLabel(f"Active dtype (Channel {self.active_chan}): {active_data.dtype}"))
+
+        # Add mode selection dropdown
+        self.mode_selector = QComboBox()
+        self.mode_selector.addItems(["8bit int", "16bit int", "32bit int", "32bit float", "64bit float"])
+        self.mode_selector.setCurrentIndex(0)  # Default to Mode 1
+        layout.addRow("Change to?:", self.mode_selector)
+
+        # Add Run button
+        run_button = QPushButton("Run")
+        run_button.clicked.connect(lambda: self.run_type(active_data))
+        layout.addRow(run_button)
+
+    def run_type(self, active_data):
+
+        mode = self.mode_selector.currentIndex()
+
+        if mode == 0:
+
+            active_data = active_data.astype(np.uint8)
+
+        elif mode == 1:
+
+            active_data = active_data.astype(np.uint16)
+
+        elif mode == 2:
+
+            active_data = active_data.astype(np.uint32)
+
+        elif mode == 3:
+
+            active_data = active_data.astype(np.float32)
+
+        elif mode == 4:
+
+            active_data = active_data.astype(np.float64)
+
+        self.parent().load_channel(self.active_chan, active_data, True)
+
+
+        print(f"Channel {self.active_chan}) dtype now: {self.parent().channel_data[self.active_chan].dtype}")
+        self.accept()
+
+
+
 
 class SkeletonizeDialog(QDialog):
     def __init__(self, parent=None):

{nettracer3d-0.6.2 → nettracer3d-0.6.3}/src/nettracer3d/segmenter.py

@@ -1289,6 +1289,9 @@ class InteractiveSegmenter:
         """Segment volume using parallel processing of chunks with vectorized chunk creation"""
         #Change the above chunk size to None to have it auto-compute largest chunks (not sure which is faster, 64 seems reasonable in test cases)
 
+        self.realtimechunks = None # Presumably no longer need this.
+        self.map_slice = None
+
         if self.mem_lock:
             chunk_size = 32 #memory efficient chunk
 
@@ -1302,7 +1305,7 @@ class InteractiveSegmenter:
             Returns:
                 List of chunks, where each chunk contains the coordinates for one z-slice or subchunk
             """
-            MAX_CHUNK_SIZE = 32768000
+            MAX_CHUNK_SIZE = 32768
             chunks = []
             
             for z in range(self.image_3d.shape[0]):
@@ -1690,6 +1693,7 @@ class InteractiveSegmenter:
     def train_batch(self, foreground_array, speed = True, use_gpu = False, use_two = False, mem_lock = False):
         """Train directly on foreground and background arrays"""
 
+        print("Training model...")
         self.speed = speed
         self.cur_gpu = use_gpu
         if mem_lock != self.mem_lock:

{nettracer3d-0.6.2 → nettracer3d-0.6.3/src/nettracer3d.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.4
 Name: nettracer3d
-Version: 0.6.2
+Version: 0.6.3
 Summary: Scripts for intializing and analyzing networks from segmentations of three dimensional images.
 Author-email: Liam McLaughlin <mclaughlinliam99@gmail.com>
 Project-URL: User_Tutorial, https://www.youtube.com/watch?v=cRatn5VTWDY
@@ -33,6 +33,7 @@ Provides-Extra: cuda12
 Requires-Dist: cupy-cuda12x; extra == "cuda12"
 Provides-Extra: cupy
 Requires-Dist: cupy; extra == "cupy"
+Dynamic: license-file
 
 NetTracer3D is a python package developed for both 2D and 3D analysis of microscopic images in the .tif file format. It supports generation of 3D networks showing the relationships between objects (or nodes) in three dimensional space, either based on their own proximity or connectivity via connecting objects such as nerves or blood vessels. In addition to these functionalities are several advanced 3D data processing algorithms, such as labeling of branched structures or abstraction of branched structures into networks. Note that nettracer3d uses segmented data, which can be segmented from other softwares such as ImageJ and imported into NetTracer3D, although it does offer its own segmentation via intensity and volumetric thresholding, or random forest machine learning segmentation. NetTracer3D currently has a fully functional GUI. To use the GUI, after installing the nettracer3d package via pip, enter the command 'nettracer3d' in your command prompt:
 
@@ -44,8 +45,10 @@ NetTracer3D is free to use/fork for academic/nonprofit use so long as citation i
 
 NetTracer3D was developed by Liam McLaughlin while working under Dr. Sanjay Jain at Washington University School of Medicine.
 
--- Version 0.6.2 updates --
+-- Version 0.6.3 updates --
 
-1. Fixed bug with performing 2D distance transforms on CPU
+1. Fixed bug with the active channel indicator on the GUI not always updating when the active channel was changed by internal loading.
 
-2. Updated ram_lock mode in the segmenter to use smaller chunks and garbage collect better.
+2. Updated ram_lock mode in the segmenter to garbage collect better... again.
+
+3. Added new function (Label neighborhoods). Find it in process -> image -> label neighborhoods. Allows a 3d labeled array to act as a kernel and extend its labels along a secondary binary array. This can be useful in evaluating nearest neighbors en-masse. The method behind this (smart label) was previously used internally for some methods (ie watershedding) but I thought I'd include it as an actual function since it has some more general uses.