napari-musa 1.0.0__py3-none-any.whl

napari_musa/Widgets_NMF.py

@@ -0,0 +1,265 @@
+""" """
+
+import sys
+from os.path import dirname
+
+sys.path.append(dirname(dirname(__file__)))
+import napari
+import numpy as np
+from magicgui.widgets import (
+    CheckBox,
+    ComboBox,
+    Container,
+    PushButton,
+    Select,
+    SpinBox,
+)
+from matplotlib.backends.backend_qt5agg import (
+    FigureCanvasQTAgg as FigureCanvas,
+)
+from matplotlib.backends.backend_qt5agg import (
+    NavigationToolbar2QT as NavigationToolbar,
+)
+from matplotlib.figure import Figure
+from napari.utils.notifications import show_info
+from qtpy.QtWidgets import (
+    QFileDialog,
+    QGroupBox,
+    QHBoxLayout,
+    QScrollArea,
+    QVBoxLayout,
+    QWidget,
+)
+from scipy.io import savemat
+
+from napari_musa.modules.functions import (
+    NMF_analysis,
+    inverse_metrics,
+)
+
+
+class NMF(QWidget):
+    """ """
+
+    def __init__(self, viewer: napari.Viewer, data, plot):
+        """ """
+        super().__init__()
+        self.viewer = viewer
+        self.data = data
+        self.plot = plot
+        self.init_ui()
+
+    def init_ui(self):
+        """ """
+        scroll = QScrollArea()
+        scroll.setWidgetResizable(True)
+        content_widget = QWidget()
+        content_layout = QVBoxLayout(content_widget)
+
+        self.build_nmf_group(content_layout)
+        content_layout.addStretch()
+
+        scroll.setWidget(content_widget)
+        main_layout = QVBoxLayout(self)
+        main_layout.addWidget(scroll)
+        self.setLayout(main_layout)
+
+    def build_nmf_group(self, layout):
+        """ """
+
+        layout.addWidget(self.create_nmf_controls())
+        layout.addWidget(self.create_mean_spectrum_area())
+        layout.addStretch()
+
+    def create_nmf_controls(self):
+        """ """
+        nmf_box = QGroupBox("NMF parameters")
+        layout = QVBoxLayout()
+        layout.addSpacing(10)
+
+        row1 = QHBoxLayout()
+        self.reduced_dataset = CheckBox(text="Apply to reduced dataset")
+        self.masked_dataset = CheckBox(text="Apply to masked dataset")
+        self.modes_combobox = ComboBox(
+            choices=self.data.modes, label="Select the imaging mode"
+        )
+        row1.addWidget(self.reduced_dataset.native)
+        row1.addWidget(self.modes_combobox.native)
+        layout.addLayout(row1)
+
+        row2 = QHBoxLayout()
+        row2.addWidget(self.masked_dataset.native)
+        layout.addLayout(row2)
+
+        self.init_dropdown = ComboBox(
+            choices=[
+                "random",
+                "nndsvd",
+                "nndsvda",
+                "nndsvdar",
+            ],
+            label="Select the initialization",
+        )
+        self.n_components = SpinBox(
+            min=1, max=500, value=5, step=1, name="N Components"
+        )
+
+        layout.addWidget(
+            Container(
+                widgets=[
+                    self.init_dropdown,
+                    self.n_components,
+                ]
+            ).native
+        )
+
+        run_btn = PushButton(text="Run NMF")
+        run_btn.clicked.connect(self.run_nmf)
+        layout.addWidget(run_btn.native)
+
+        row3 = QHBoxLayout()
+        self.nmf_basis_multiselecton = Select(label="Select Bases", choices=[])
+        self.nmf_basis_multiselecton.changed.connect(
+            self.on_basis_selection_changed
+        )
+        row3.addWidget(self.nmf_basis_multiselecton.native)
+        layout.addLayout(row3)
+        nmf_box.setLayout(layout)
+        return nmf_box
+
+    def create_mean_spectrum_area(self):
+        """ """
+        nmf_box = QGroupBox("NMF spectra")
+        layout = QVBoxLayout()
+        layout.addSpacing(10)
+        self.mean_plot = FigureCanvas(Figure(figsize=(5, 3)))
+        self.mean_plot.setMinimumSize(300, 450)
+        self.mean_plot_toolbar = NavigationToolbar(self.mean_plot, self)
+        self.plot.customize_toolbar(self.mean_plot_toolbar)
+        self.plot.setup_plot(self.mean_plot)
+
+        layout.addWidget(self.mean_plot)
+        layout.addWidget(self.mean_plot_toolbar)
+
+        # Export button
+        export_btn = PushButton(text="Export spectra")
+        export_btn.clicked.connect(self.export_spectrum)
+        export_nmf_btn = PushButton(text="Export NMF matrices as .mat")
+        export_nmf_btn.clicked.connect(self.export_nmf)
+
+        layout.addWidget(Container(widgets=[export_btn]).native)
+        layout.addWidget(Container(widgets=[export_nmf_btn]).native)
+        nmf_box.setLayout(layout)
+        return nmf_box
+
+    def run_nmf(self):
+        """Perform NMF"""
+        mode = self.modes_combobox.value
+
+        n_basis = self.n_components.value
+        options = [f"Basis {i}" for i in range(n_basis)]
+
+        if self.masked_dataset.value:
+            dataset = self.data.hypercubes_masked[mode]
+            data_reshaped = dataset.reshape(
+                dataset.shape[0] * dataset.shape[1], -1
+            )
+            self.points = np.array(
+                np.where(~np.isnan(np.mean(data_reshaped, axis=1)))
+            ).flatten()
+            print(self.points)
+
+        elif self.reduced_dataset.value:
+            dataset = self.data.hypercubes_red[mode]
+            self.points = []
+        else:
+            dataset = self.data.hypercubes[mode]
+            self.points = []
+
+        self.data.nmf_maps[mode], self.data.nmf_basis[mode] = NMF_analysis(
+            dataset,
+            points=self.points,
+            n_components=n_basis,
+            init=self.init_dropdown.value,
+        )
+
+        self.nmf_basis_multiselecton.choices = options
+
+        self.viewer.add_image(
+            self.data.nmf_maps[mode].transpose(2, 0, 1),
+            name=str(mode) + " - NMF",
+            # ={"type": "hyperspectral_cube"},
+        )
+
+        show_info("NMF analysis completed!")
+
+    def on_basis_selection_changed(self, value):
+        mode = self.modes_combobox.value
+
+        print("Selected bases:", value)
+        print("Shape of the array:", self.data.nmf_basis[mode].shape)
+
+        self.basis_numbers = sorted([int(s.split()[1]) for s in value])
+        self.selected_basis = self.data.nmf_basis[mode][:, self.basis_numbers]
+        print(self.selected_basis.shape)
+        self.selected_basis_to_show = self.selected_basis
+
+        if mode == "Fused":
+            fusion_point = self.data.wls[self.data.fusion_modes[0]].shape[0]
+            print(self.data.fusion_norm)
+            # xxx aggiustare corrections
+            self.selected_basis_to_show[:fusion_point, :] = inverse_metrics(
+                self.selected_basis_to_show[:fusion_point, :],
+                self.data.fusion_norm,
+                self.data.fusion_params[0],
+            )
+            self.selected_basis_to_show[fusion_point:, :] = inverse_metrics(
+                self.selected_basis_to_show[fusion_point:, :],
+                self.data.fusion_norm,
+                self.data.fusion_params[1],
+            )
+
+        self.plot.show_spectra(
+            self.mean_plot,
+            self.selected_basis_to_show,
+            mode,
+            basis_numbers=self.basis_numbers,
+            export_txt_flag=False,
+        )
+
+        # show_info(f"NMF bases selected: {self.basis_numbers}")
+
+    def export_spectrum(self):
+        """Export the mean spectrum"""
+        mode = self.modes_combobox.value
+
+        self.plot.show_spectra(
+            self.mean_plot,
+            self.selected_basis_to_show,
+            mode,
+            basis_numbers=self.basis_numbers,
+            export_txt_flag=True,
+        )
+
+    def export_nmf(self):
+        """Export nmf"""
+        mode = self.modes_combobox.value
+        H = self.data.nmf_maps[mode]
+        W = self.data.nmf_basis[mode]
+
+        save_dict = {
+            "H": H,
+            "W": W,
+        }
+
+        filename, _ = QFileDialog.getSaveFileName(
+            self, "Save nmf .mat", "", "mat (*.mat)"
+        )
+        if filename:
+            savemat(filename, save_dict)
+
+    def update_number_H(self):
+        """ """
+        index = self.viewer.dims.current_step[0]
+        index = min(index, self.n_components.value - 1)
+        self.viewer.text_overlay.text = f"Component number: {index}"

napari_musa/Widgets_PCA.py

@@ -0,0 +1,212 @@
+""" """
+
+import sys
+from os.path import dirname
+
+sys.path.append(dirname(dirname(__file__)))
+import napari
+import numpy as np
+import pyqtgraph as pg
+from magicgui.widgets import (
+    CheckBox,
+    ComboBox,
+    Container,
+    PushButton,
+    SpinBox,
+)
+from qtpy.QtWidgets import (
+    QGroupBox,
+    QHBoxLayout,
+    QScrollArea,
+    QVBoxLayout,
+    QWidget,
+)
+
+from napari_musa.modules.functions import PCA_analysis, RGB_to_hex
+
+
+class PCA(QWidget):
+    def __init__(self, viewer: napari.Viewer, data, plot):
+        super().__init__()
+        self.viewer = viewer
+        self.data = data
+        self.plot = plot
+        self.hex_reshaped = np.zeros(1)
+        self.init_ui()
+
+    def init_ui(self):
+        """ """
+        scroll = QScrollArea()
+        scroll.setWidgetResizable(True)
+        content_widget = QWidget()
+        content_layout = QVBoxLayout(content_widget)
+
+        self.build_sivm_group(content_layout)
+        content_layout.addStretch()
+
+        scroll.setWidget(content_widget)
+        main_layout = QVBoxLayout(self)
+        main_layout.addWidget(scroll)
+        self.setLayout(main_layout)
+
+    def build_sivm_group(self, layout):
+        """ """
+        layout.addWidget(self.create_pca_controls())
+        layout.addWidget(self.create_pca_scatterplot())
+        layout.addStretch()
+
+    def create_pca_controls(self):
+        PCA_box = QGroupBox("PCA parameters")
+        PCA_main_layout = QVBoxLayout()
+        PCA_main_layout.addSpacing(10)
+        # - - - pca data - - -
+        self.reduced_dataset = CheckBox(text="Apply to reduced dataset")
+        self.modes_combobox = ComboBox(
+            choices=self.data.modes, label="Select the imaging mode"
+        )  # DROPDOWN FOR CALIBRATION
+        self.n_components = SpinBox(
+            min=1, max=100, value=10, step=1, name="Number of components"
+        )
+
+        PCA_perform_btn = PushButton(text="Perform PCA")
+        PCA_perform_btn.clicked.connect(self.PCA_perform_btn_f)
+        PCA_main_layout.addWidget(
+            Container(
+                widgets=[
+                    self.reduced_dataset,
+                    self.modes_combobox,
+                    self.n_components,
+                    PCA_perform_btn,
+                ]
+            ).native
+        )
+        PCA_box.setLayout(PCA_main_layout)
+        return PCA_box
+
+    def create_pca_scatterplot(self):
+        PCA_box = QGroupBox("PCA scatterplot")
+        PCA_layout_plot_var = QVBoxLayout()
+        PCA_layout_plot_var.addSpacing(10)
+        # - - - pca variables - - -
+        self.x_axis = SpinBox(min=1, max=100, value=1, step=1, name="X axis")
+        self.y_axis = SpinBox(min=1, max=100, value=2, step=1, name="Y axis")
+        PCA_layout_plot_var.addWidget(
+            Container(widgets=[self.x_axis, self.y_axis]).native
+        )
+        PCA_layout_perform = QHBoxLayout()
+        self.PCA_colorRGB = CheckBox(text="Scatterplot with True RGB")
+        PCA_show_plot_btn = PushButton(text="Show PCA scatterplot")
+        PCA_show_plot_btn.clicked.connect(self.PCA_show_plot_btn_f)
+        PCA_layout_perform.addWidget(
+            Container(
+                widgets=[self.PCA_colorRGB, PCA_show_plot_btn],
+                layout="horizontal",
+            ).native
+        )
+        PCA_layout_plot_var.addLayout(PCA_layout_perform)
+
+        self.pca_plot = pg.PlotWidget()
+        self.plot.setup_scatterplot(self.pca_plot)
+
+        # Add control buttons for scatter plot interaction
+        btn_layout = QHBoxLayout()
+
+        for icon, func in [
+            ("fa5s.home", lambda: self.pca_plot.getViewBox().autoRange()),
+            (
+                "fa5s.draw-polygon",
+                lambda: self.plot.polygon_selection(self.pca_plot),
+            ),
+            ("ri.add-box-fill", self.handle_selection),
+            (
+                "mdi6.image-edit",
+                lambda: self.plot.save_image_button(self.pca_plot),
+            ),
+        ]:
+            btn = self.plot.create_button(icon)
+            btn.clicked.connect(func)
+            btn_layout.addWidget(btn)
+
+        self.point_size = SpinBox(
+            min=1, max=100, value=1, step=1, name="Point size"
+        )
+        btn_layout.addSpacing(30)
+        btn_layout.addWidget(Container(widgets=[self.point_size]).native)
+        PCA_layout_plot_var.addLayout(btn_layout)
+        PCA_layout_plot_var.addWidget(self.pca_plot)
+        PCA_box.setLayout(PCA_layout_plot_var)
+        return PCA_box
+
+    def PCA_perform_btn_f(self):
+        mode = self.modes_combobox.value
+        if self.reduced_dataset.value:
+            self.PCA_dataset = self.data.hypercubes_red[mode]
+        else:
+            self.PCA_dataset = self.data.hypercubes[mode]
+
+        self.data.pca_maps[mode], W = PCA_analysis(
+            self.PCA_dataset, self.n_components.value
+        )
+
+        print("PCA dataset shape: ", self.data.pca_maps[mode].shape)
+        # Add the PCA maps to the viewer
+        self.viewer.add_image(
+            self.data.pca_maps[mode].transpose(2, 0, 1),
+            name=str(mode) + " - PCA ",
+            colormap="gray_r",
+            # ={"type": "hyperspectral_cube"},
+        )
+
+    def PCA_show_plot_btn_f(self):
+        """Plot UMAP scatter plot"""
+        mode = self.modes_combobox.value
+        pca_xaxis = self.x_axis.value - 1
+        pca_yaxis = self.y_axis.value - 1
+        H_PCA_reshaped = self.data.pca_maps[self.modes_combobox.value].reshape(
+            -1, self.n_components.value
+        )
+        self.H_PCA_reshaped_selected = np.stack(
+            (H_PCA_reshaped[:, pca_xaxis], H_PCA_reshaped[:, pca_yaxis])
+        ).T
+
+        if self.PCA_colorRGB.value:
+            if self.reduced_dataset.value:
+                colors = np.array(RGB_to_hex(self.data.rgb_red[mode])).reshape(
+                    -1
+                )
+            else:
+                colors = np.array(RGB_to_hex(self.data.rgb[mode])).reshape(-1)
+        else:
+            colors = pg.mkBrush("#262930")
+
+        # print("Colors: \n", colors.reshape(-1))
+        self.points = []
+        self.plot.show_scatterplot(
+            self.pca_plot,
+            self.H_PCA_reshaped_selected,
+            colors,
+            self.points,
+            self.point_size.value,
+        )
+
+    def handle_selection(self):
+        """Handle polygon selection and create label layer"""
+        mode = self.modes_combobox.value
+        if self.reduced_dataset.value:
+            dataset = self.data.hypercubes_red[mode]
+            self.points = []
+
+        else:
+            dataset = self.data.hypercubes[mode]
+        self.plot.show_selected_points(
+            self.H_PCA_reshaped_selected,
+            dataset,
+            mode,
+            self.points,
+        )
+
+    def update_number_H(self):
+        """ """
+        index = self.viewer.dims.current_step[0]
+        index = min(index, self.n_components.value - 1)
+        self.viewer.text_overlay.text = f"PCA number: {index}"