AOT-biomaps 2.9.167__py3-none-any.whl → 2.9.273__py3-none-any.whl

AOT_biomaps/AOT_Acoustic/StructuredWave.py

@@ -292,8 +292,8 @@ class StructuredWave(AcousticField):
                 f"!number format := short float\n"
                 f"!number of bytes per pixel := 4\n"
                 f"scaling factor (mm/pixel) [1] := {self.params['dx'] * 1000}\n"
-                f"scaling factor (mm/pixel) [2] := {self.params['dx'] * 1000}\n"
-                f"scaling factor (s/pixel) [3] := {1 / self.params['f_AQ']}\n"
+                f"scaling factor (mm/pixel) [2] := {self.params['dz'] * 1000}\n"
+                f"scaling factor (s/pixel) [3] := {1 / self.params['f_saving']}\n"
                 f"first pixel offset (mm) [1] := {self.params['Xrange'][0] * 1e3}\n"
                 f"first pixel offset (mm) [2] := {self.params['Zrange'][0] * 1e3}\n"
                 f"first pixel offset (s) [3] := 0\n"

AOT_biomaps/AOT_Acoustic/_mainAcoustic.py

@@ -19,10 +19,12 @@ from kwave.kspaceFirstOrder3D import kspaceFirstOrder3D
 from kwave.kspaceFirstOrder2D import kspaceFirstOrder2D
 from kwave.options.simulation_options import SimulationOptions
 from kwave.options.simulation_execution_options import SimulationExecutionOptions
+from AOT_biomaps.Settings import Params
 
 from tempfile import gettempdir
 from math import ceil
 from abc import ABC, abstractmethod
+import logging
 
 
 
@@ -75,7 +77,7 @@ class AcousticField(ABC):
             print(f"Initialization error: {e}")
             raise
         if params != None:
-            if type(params) != AOT_biomaps.Settings.Params:
+            if type(params) != Params:
                 raise TypeError("params must be an instance of the Params class")
 
             self.params = {
@@ -225,6 +227,7 @@ class AcousticField(ABC):
         Generate the acoustic field based on the specified simulation type and parameters.
         """
         try:
+            logging.getLogger('root').setLevel(logging.ERROR)
             if self.params['typeSim'] == TypeSim.FIELD2.value:
                 raise NotImplementedError("FIELD2 simulation is not implemented yet.")
             elif self.params['typeSim'] == TypeSim.KWAVE.value:
@@ -233,7 +236,7 @@ class AcousticField(ABC):
                         field = self._generate_acoustic_field_KWAVE_2D(isGpu, show_log)
                     except Exception as e:
                         raise RuntimeError(f"Failed to generate 2D acoustic field: {e}")
-                    self.field = calculate_envelope_squared(field)
+                    self.field = reshape_field(calculate_envelope_squared(field),[self.factorT, self.factorX, self.factorZ])
                 elif self.params["dim"] == Dim.D3.value:
                     field = self._generate_acoustic_field_KWAVE_3D(isGpu, show_log)
                     self.field = reshape_field(calculate_envelope_squared(field),[self.factorT, self.factorX, self.factorZ])
@@ -265,7 +268,7 @@ class AcousticField(ABC):
             print(f"Error in save_field method: {e}")
             raise
 
-    def load_field(self, folderPath, formatSave=FormatSave.HDR_IMG):
+    def load_field(self, folderPath, formatSave=FormatSave.HDR_IMG, nameBlock=None):
         """
         Load the acoustic field from a file in the specified format.
 
@@ -286,7 +289,7 @@ class AcousticField(ABC):
                         raise NotImplementedError("3D KWAVE field loading is not implemented yet.")
                 elif formatSave.value == FormatSave.H5.value:
                     if self.params["dim"] == Dim.D2.value:
-                         self._load_field_h5(folderPath)
+                         self._load_field_h5(folderPath,nameBlock)
                     elif self.params["dim"] == Dim.D3.value:
                         raise NotImplementedError("H5 KWAVE field loading is not implemented yet.")
                 elif formatSave.value == FormatSave.NPY.value:
@@ -531,6 +534,8 @@ class AcousticField(ABC):
             total_size_z = next_power_of_2(Nz)
             pml_x_size = (total_size_x - Nx) // 2
             pml_z_size = (total_size_z - Nz) // 2
+            pml_x_size = max(pml_x_size, 50)    # Ensure a minimum PML size of 50 grid points to avoid parasitic reflections
+            pml_z_size = max(pml_z_size, 50)    # Ensure a minimum PML size of 50 grid points to avoid parasitic reflections
 
             # --- 6. Simulation options ---
             simulation_options = SimulationOptions(
@@ -650,7 +655,7 @@ class AcousticField(ABC):
         """
         pass
 
-    def _load_field_h5(self, filePath):
+    def _load_field_h5(self, filePath,nameBlock):
         """
         Load the 2D acoustic field from an H5 file.
 
@@ -661,8 +666,10 @@ class AcousticField(ABC):
         - field (numpy.ndarray): The loaded acoustic field.
         """
         try:
-            with h5py.File(filePath+self.getName_field()+".h5", 'r') as f:
-                self.field = f['data'][:]
+            if nameBlock is None:
+                nameBlock = 'data'
+            with h5py.File(os.path.join(filePath, self.getName_field()+".h5"), 'r') as f:
+                self.field = f[nameBlock][:]
         except Exception as e:
             print(f"Error in _load_field_h5 method: {e}")
             raise

AOT_biomaps/AOT_Experiment/Tomography.py

@@ -7,6 +7,8 @@ import psutil
 import numpy as np
 import matplotlib.pyplot as plt
 from tqdm import trange
+import h5py
+from scipy.io import loadmat
 
 class Tomography(Experiment):
     def __init__(self, **kwargs):
@@ -47,7 +49,7 @@ class Tomography(Experiment):
             return False, f"OpticImage phantom shape {self.OpticImage.phantom.shape} does not match AcousticFields shape {self.AcousticFields[0].field.shape[1:]}."
         return True, "Experiment is correctly initialized."
 
-    def generateAcousticFields(self, fieldDataPath=None, show_log=True):
+    def generateAcousticFields(self, fieldDataPath=None, show_log=True, nameBlock=None):
         """
         Generate the acoustic fields for simulation.
         Args:
@@ -57,7 +59,7 @@ class Tomography(Experiment):
             systemMatrix: A numpy array of the generated fields.
         """
         if self.TypeAcoustic.value == WaveType.StructuredWave.value:
-            self.AcousticFields = self._generateAcousticFields_STRUCT_CPU(fieldDataPath, show_log)
+            self.AcousticFields = self._generateAcousticFields_STRUCT_CPU(fieldDataPath, show_log, nameBlock)
         else:
             raise ValueError("Unsupported wave type.")
 
@@ -389,7 +391,7 @@ class Tomography(Experiment):
         return True
 
     # PRIVATE METHODS
-    def _generateAcousticFields_STRUCT_CPU(self, fieldDataPath=None, show_log=False):
+    def _generateAcousticFields_STRUCT_CPU(self, fieldDataPath=None, show_log=False, nameBlock=None):
         if self.patterns is None:
             raise ValueError("patterns is not initialized. Please load or generate the active list first.")
         listAcousticFields = []
@@ -415,7 +417,7 @@ class Tomography(Experiment):
             if pathField is not None and os.path.exists(pathField):
                 progress_bar.set_postfix_str(f"Loading field - {AcousticField.getName_field()} -- Memory used: {memory.percent}%")
                 try:
-                    AcousticField.load_field(fieldDataPath, self.FormatSave)
+                    AcousticField.load_field(fieldDataPath, self.FormatSave,nameBlock)
                 except:
                     progress_bar.set_postfix_str(f"Error loading field -> Generating field - {AcousticField.getName_field()} -- Memory used: {memory.percent}% ---- processing on {config.get_process().upper()} ----")
                     AcousticField.generate_field(show_log=show_log)
@@ -433,3 +435,71 @@ class Tomography(Experiment):
             listAcousticFields.append(AcousticField)
             progress_bar.set_postfix_str("")
         return listAcousticFields
+
+    def load_experimentalAO(self, pathAO, withTumor = True, h5name='AOsignal'):
+        """
+        Load experimental AO signals from specified file paths.
+        Args:
+            path_withTumor: Path to the AO signal with tumor.
+            path_withoutTumor: Path to the AO signal without tumor.
+        """
+        if not os.path.exists(pathAO):
+            raise FileNotFoundError(f"File {pathAO} not found.")
+
+        if pathAO.endswith('.npy'):
+            ao_signal = np.load(pathAO)
+        elif pathAO.endswith('.h5'):
+            with h5py.File(pathAO, 'r') as f:
+                if h5name not in f:
+                    raise KeyError(f"Dataset '{h5name}' not found in the HDF5 file.")
+                ao_signal = f[h5name][:]
+        elif pathAO.endswith('.mat'):
+            mat_data = loadmat(pathAO)
+            if h5name not in mat_data:
+                raise KeyError(f"Dataset '{h5name}' not found in the .mat file.")
+            ao_signal = mat_data[h5name]
+        elif pathAO.endswith('.hdr'):
+            ao_signal = self._loadAOSignal(pathAO)
+        else:
+            raise ValueError("Unsupported file format. Supported formats are: .npy, .h5, .mat, .hdr")
+        
+        if withTumor:
+            self.AOsignal_withTumor = ao_signal
+        else:
+            self.AOsignal_withoutTumor = ao_signal
+
+    def check_experimentalAO(self, activeListPath, withTumor=True):
+        """
+        Check if the experimental AO signals are correctly initialized.
+        """
+        if withTumor:
+            if self.AOsignal_withTumor is None:
+                raise ValueError("Experimental AOsignal with tumor is not initialized. Please load the experimental AO signal with tumor first.")
+        else:
+            if self.AOsignal_withoutTumor is None:
+                raise ValueError("Experimental AOsignal without tumor is not initialized. Please load the experimental AO signal without tumor first.")
+        if self.AcousticFields is not None:
+            # get min time shape between all AO signals
+            print()
+            
+            if self.AcousticFields[0].field.shape[0] > self.AOsignal_withTumor.shape[0]:
+                self.cutAcousticFields(max_t=self.AOsignal_withTumor.shape[0]/float(self.params.acoustic['f_saving']))
+            else:
+                for i in range(len(self.AcousticFields)):
+                    min_time_shape = min(self.AcousticFields[i].field.shape[0])
+                if withTumor:
+                    self.AOsignal_withTumor = self.AOsignal_withTumor[:min_time_shape, :]
+                else:
+                    self.AOsignal_withoutTumor = self.AOsignal_withoutTumor[:min_time_shape, :]
+
+            for field in self.AcousticFields:                        
+                if activeListPath is not None:
+                    with open(activeListPath, 'r') as file:
+                        lines = file.readlines()
+                        expected_name = lines[self.AcousticFields.index(field)].strip()
+                        nameField = field.getName_field()
+                        if nameField.startswith("field_"):
+                            nameField = nameField[len("field_"):]
+                        if nameField != expected_name:
+                            raise ValueError(f"Field name {nameField} does not match the expected name {expected_name} from the active list.")
+        print("Experimental AO signals are correctly initialized.")

AOT_biomaps/AOT_Experiment/_mainExperiment.py

@@ -56,6 +56,7 @@ class Experiment(ABC):
         pass
 
     def cutAcousticFields(self, max_t, min_t=0):
+
         max_t = float(max_t)
         min_t = float(min_t)
 
@@ -113,7 +114,6 @@ class Experiment(ABC):
             noiseSignals[:, i] = noisy_signal
         return noiseSignals
 
-
     def reduceDims(self, mode='avg'):
         """
         Réduit les dimensions T, X, Z d'un numpy array (T, X, Z) par 2 en utilisant une convolution.
@@ -154,6 +154,16 @@ class Experiment(ABC):
         for param in ['dx', 'dy', 'dz']:
             convert_and_update(self.params.general, param, lambda x: x * 2)
 
+    def normalizeAOsignals(self, withTumor=True):
+        if withTumor and self.AOsignal_withTumor is None:
+            raise ValueError("AO signal with tumor is not generated. Please generate it first.")
+        if not withTumor and self.AOsignal_withoutTumor is None:
+            raise ValueError("AO signal without tumor is not generated. Please generate it first.")
+        if withTumor:
+            self.AOsignal_withTumor = self.AOsignal_withTumor - np.min(self.AOsignal_withTumor)/(np.max(self.AOsignal_withTumor)-np.min(self.AOsignal_withTumor))
+        else:
+            self.AOsignal_withoutTumor = self.AOsignal_withoutTumor - np.min(self.AOsignal_withoutTumor)/(np.max(self.AOsignal_withoutTumor)-np.min(self.AOsignal_withoutTumor))
+
     def saveAcousticFields(self, save_directory):
         progress_bar = trange(len(self.AcousticFields), desc="Saving Acoustic Fields")
         for i in progress_bar:
@@ -230,67 +240,97 @@ class Experiment(ABC):
         return ani
 
     def generateAOsignal(self, withTumor=True, AOsignalDataPath=None):
-        if self.AcousticFields is None:
-            raise ValueError("AcousticFields is not initialized. Please generate the system matrix first.")
-
-        if self.OpticImage is None:
-            raise ValueError("OpticImage is not initialized. Please generate the phantom first.")
 
         if AOsignalDataPath is not None:
             if not os.path.exists(AOsignalDataPath):
                 raise FileNotFoundError(f"AO file {AOsignalDataPath} not found.")
-            AOmatrix = self._load_AOSignal(AOsignalDataPath)
-            if AOmatrix.shape[0] != self.AcousticFields[0].field.shape[0]:
-                print(f"AO signal shape {AOmatrix.shape} does not match the expected shape {self.AcousticFields[0].field.shape}. Generating corrected AO signal to match...")
+            if withTumor:
+                self.AOsignal_withTumor = self._loadAOSignal(AOsignalDataPath)
+                if self.AOsignal_withTumor.shape[0] != self.AcousticFields[0].field.shape[0]:
+                    print(f"AO signal shape {self.AOsignal_withTumor.shape} does not match the expected shape {self.AcousticFields[0].field.shape}. Resizing Acoustic fields...")
+                    self.cutAcousticFields(max_t=self.AOsignal_withTumor.shape[0] / float(self.params.acoustic['f_saving']), min_t=0)
             else:
-                return AOmatrix
-
-        if not all(field.field.shape == self.AcousticFields[0].field.shape for field in self.AcousticFields):
-            minShape = min([field.field.shape[0] for field in self.AcousticFields])
-            self.cutAcousticFields(minShape * self.params['fs_aq'])
-        else:
-            shape_field = self.AcousticFields[0].field.shape
-
-        AOsignal = np.zeros((shape_field[0], len(self.AcousticFields)), dtype=np.float32)
-
-        if withTumor:
-            description = "Generating AO Signal with Tumor"
-        else:
-            description = "Generating AO Signal without Tumor"
+                self.AOsignal_withoutTumor = self._loadAOSignal(AOsignalDataPath)
+                if self.AOsignal_withoutTumor.shape[0] != self.AcousticFields[0].field.shape[0]:
+                    print(f"AO signal shape {self.AOsignal_withoutTumor.shape} does not match the expected shape {self.AcousticFields[0].field.shape}. Resizing Acoustic fields...")
+                    self.cutAcousticFields(max_t=self.AOsignal_withoutTumor.shape[0] / float(self.params.acoustic['f_saving']), min_t=0)
+        else:    
+            if self.AcousticFields is None:
+                raise ValueError("AcousticFields is not initialized. Please generate the system matrix first.")
+
+            if self.OpticImage is None:
+                raise ValueError("OpticImage is not initialized. Please generate the phantom first.")
+            
+            if not all(field.field.shape == self.AcousticFields[0].field.shape for field in self.AcousticFields):
+                minShape = min([field.field.shape[0] for field in self.AcousticFields])
+                self.cutAcousticFields(max_t=minShape * self.params['fs_aq'])
+            else:
+                shape_field = self.AcousticFields[0].field.shape
 
-        for i in trange(len(self.AcousticFields), desc=description):
-            for t in range(self.AcousticFields[i].field.shape[0]):
-                if withTumor:
-                    interaction = self.OpticImage.phantom * self.AcousticFields[i].field[t, :, :]
-                else:
-                    interaction = self.OpticImage.laser.intensity * self.AcousticFields[i].field[t, :, :]
-                AOsignal[t, i] = np.sum(interaction)
+            AOsignal = np.zeros((shape_field[0], len(self.AcousticFields)), dtype=np.float32)
 
-        if withTumor:
-            self.AOsignal_withTumor = AOsignal
-        else:
-            self.AOsignal_withoutTumor = AOsignal
+            if withTumor:
+                description = "Generating AO Signal with Tumor"
+            else:
+                description = "Generating AO Signal without Tumor"
+
+            for i in trange(len(self.AcousticFields), desc=description):
+                for t in range(self.AcousticFields[i].field.shape[0]):
+                    if withTumor:
+                        interaction = self.OpticImage.phantom * self.AcousticFields[i].field[t, :, :]
+                    else:
+                        interaction = self.OpticImage.laser.intensity * self.AcousticFields[i].field[t, :, :]
+                    AOsignal[t, i] = np.sum(interaction)
+
+            if withTumor:
+                self.AOsignal_withTumor = AOsignal
+            else:
+                self.AOsignal_withoutTumor = AOsignal
 
     @staticmethod
-    def _loadAOSignal(cdh_file):
-        with open(cdh_file, "r") as file:
-            cdh_content = file.readlines()
-
-        n_events = int([line.split(":")[1].strip() for line in cdh_content if "Number of events" in line][0])
-        n_acquisitions = int([line.split(":")[1].strip() for line in cdh_content if "Number of acquisitions per event" in line][0])
-
-        AOsignal_matrix = np.zeros((n_events, n_acquisitions), dtype=np.float32)
-
-        with open(cdh_file.replace(".cdh", ".cdf"), "rb") as file:
-            for event in range(n_events):
-                num_elements = int([line.split(":")[1].strip() for line in cdh_content if "Number of US transducers" in line][0])
-                hex_length = (num_elements + 3) // 4
-                file.read(hex_length // 2)
-
-                signal = np.frombuffer(file.read(n_acquisitions * 4), dtype=np.float32)
-                AOsignal_matrix[event, :] = signal
-
-        return AOsignal_matrix
+    def _loadAOSignal(AOsignalPath):
+        if AOsignalPath.endswith(".cdh"):
+            with open(AOsignalPath, "r") as file:
+                cdh_content = file.readlines()
+            
+            cdf_path = AOsignalPath.replace(".cdh", ".cdf")
+
+            # Extraire les paramètres depuis le fichier .cdh
+            n_scans = int([line.split(":")[1].strip() for line in cdh_content if "Number of events" in line][0])
+            n_acquisitions_per_event = int([line.split(":")[1].strip() for line in cdh_content if "Number of acquisitions per event" in line][0])
+            num_elements = int([line.split(":")[1].strip() for line in cdh_content if "Number of US transducers" in line][0])
+
+            # Initialisation des structures
+            AO_signal = np.zeros((n_acquisitions_per_event, n_scans), dtype=np.float32)
+            active_lists = []
+            angles = []
+
+            # Lecture du fichier binaire
+            with open(cdf_path, "rb") as file:
+                for j in trange(n_scans, desc="Lecture des événements"):
+                    # Lire l'activeList : 48 caractères hex = 24 bytes
+                    active_list_bytes = file.read(24)
+                    active_list_hex = active_list_bytes.hex()
+                    active_lists.append(active_list_hex)
+
+                    # Lire l'angle (1 byte signé)
+                    angle_byte = file.read(1)
+                    angle = np.frombuffer(angle_byte, dtype=np.int8)[0]
+                    angles.append(angle)
+
+                    # Lire le signal AO (float32)
+                    data = np.frombuffer(file.read(n_acquisitions_per_event * 4), dtype=np.float32)
+                    if len(data) != n_acquisitions_per_event:
+                        raise ValueError(f"Erreur à l'événement {j} : attendu {n_acquisitions_per_event}, obtenu {len(data)}")
+                    AO_signal[:, j] = data
+
+            return AO_signal
+
+
+        elif AOsignalPath.endswith(".npy"):
+            return np.load(AOsignalPath)  # Supposé déjà au bon format
+        else:
+            raise ValueError("Format de fichier non supporté. Utilisez .cdh/.cdf ou .npy.")
 
     def saveAOsignals_Castor(self, save_directory, withTumor=True):
         if withTumor:
@@ -318,10 +358,10 @@ class Experiment(ABC):
         header_content = (
             f"Data filename: {'AOSignals_withTumor.cdf' if withTumor else 'AOSignals_withoutTumor.cdf'}\n"
             f"Number of events: {nScan}\n"
-            f"Number of acquisitions per event: {AO_signal.shape[1]}\n"
+            f"Number of acquisitions per event: {AO_signal.shape[0]}\n"
             f"Start time (s): 0\n"
             f"Duration (s): 1\n"
-            f"Acquisition frequency (Hz): {1/self.AcousticFields[0].kgrid.dt}\n"
+            f"Acquisition frequency (Hz): {self.params.acoustic['f_saving']}\n"
             f"Data mode: histogram\n"
             f"Data type: AOT\n"
             f"Number of US transducers: {self.params.acoustic['num_elements']}"

AOT_biomaps/AOT_Recon/AOT_Optimizers/DEPIERRO.py

@@ -1,6 +1,8 @@
 from AOT_biomaps.AOT_Recon.ReconEnums import PotentialType
-from AOT_biomaps.AOT_Recon.ReconTools import _build_adjacency_sparse
+from AOT_biomaps.AOT_Recon.ReconTools import _build_adjacency_sparse, calculate_memory_requirement, check_gpu_memory
 from AOT_biomaps.Config import config
+
+import warnings
 import numpy as np
 import torch
 from tqdm import trange
@@ -11,9 +13,42 @@ if config.get_process() == 'gpu':
     except ImportError:
         raise ImportError("torch_scatter and torch_sparse are required for GPU processing. Please install them using 'pip install torch-scatter torch-sparse' with correct link (follow instructions https://github.com/LucasDuclos/AcoustoOpticTomography/edit/main/README.md).")
 
-def _DEPIERRO_GPU(SMatrix, y, numIterations, beta, sigma, isSavingEachIteration, withTumor, max_saves=5000):
-    # Initialisation du device
-    device = torch.device(f"cuda:{config.select_best_gpu()}")
+def DEPIERRO(
+        SMatrix,
+        y, 
+        numIterations,
+        beta,
+        sigma,
+        isSavingEachIteration,
+        withTumor, 
+        max_saves, 
+        show_logs):
+        """
+        This method implements the DEPIERRO algorithm using either CPU or single-GPU PyTorch acceleration.
+        Multi-GPU and Multi-CPU modes are not implemented for this algorithm.
+        """
+        try:
+            tumor_str = "WITH" if withTumor else "WITHOUT"
+            # Auto-select device and method
+            if device is None:
+                if torch.cuda.is_available() and check_gpu_memory(config.select_best_gpu(), calculate_memory_requirement(SMatrix, y), show_logs=show_logs):
+                    device = torch.device(f"cuda:{config.select_best_gpu()}")
+                    use_gpu = True
+                else:
+                    device = torch.device("cpu")
+                    use_gpu = False
+            else:
+                use_gpu = device.type == "cuda"
+            # Dispatch to the appropriate implementation
+            if use_gpu:
+                return _DEPIERRO_GPU(SMatrix, y, numIterations, beta, sigma, isSavingEachIteration, tumor_str, device, max_saves, show_logs)
+            else:
+                return _DEPIERRO_CPU(SMatrix, y, numIterations, beta, sigma, isSavingEachIteration, tumor_str, device, max_saves, show_logs)
+        except Exception as e:
+            print(f"Error in MLEM: {type(e).__name__}: {e}")
+            return None, None
+
+def _DEPIERRO_GPU(SMatrix, y, numIterations, beta, sigma, isSavingEachIteration, tumor_str, device, max_saves, show_logs=True):
     # Conversion des données en tenseurs PyTorch (float64)
     A_matrix_torch = torch.tensor(SMatrix, dtype=torch.float64, device=device)
     y_torch = torch.tensor(y, dtype=torch.float64, device=device)
@@ -33,7 +68,7 @@ def _DEPIERRO_GPU(SMatrix, y, numIterations, beta, sigma, isSavingEachIteration,
     # Construction de la matrice d'adjacence
     adj_index, adj_values = _build_adjacency_sparse(Z, X, device=device, dtype=torch.float64)
     # Description pour la barre de progression
-    description = f"AOT-BioMaps -- Bayesian Reconstruction Tomography: DE PIERRO (Sparse QUADRATIC β:{beta:.4f}, σ:{sigma:.4f}) ---- {'WITH' if withTumor else 'WITHOUT'} TUMOR ---- processing on single GPU no.{torch.cuda.current_device()}"
+    description = f"AOT-BioMaps -- Bayesian Reconstruction Tomography: DE PIERRO (Sparse QUADRATIC β:{beta:.4f}, σ:{sigma:.4f}) ---- {tumor_str} TUMOR ---- processing on single GPU no.{torch.cuda.current_device()}"
     # Configuration pour la sauvegarde des itérations
     saved_indices = [0]
 
@@ -47,7 +82,8 @@ def _DEPIERRO_GPU(SMatrix, y, numIterations, beta, sigma, isSavingEachIteration,
             save_indices.append(numIterations - 1)
 
     # Boucle principale MAP-EM
-    for it in trange(numIterations, desc=description):
+    iterator = trange(numIterations, desc=description) if show_logs else range(numIterations)
+    for it in iterator:
         theta_p = matrix_theta_torch[-1]
         theta_p_flat = theta_p.reshape(-1)
         # Étape 1 : Projection avant
@@ -92,10 +128,8 @@ def _DEPIERRO_GPU(SMatrix, y, numIterations, beta, sigma, isSavingEachIteration,
     else:
         return matrix_theta_torch[-1].cpu().numpy(), None
 
-def _DEPIERRO_CPU(SMatrix, y, Omega, numIterations, beta, sigma, isSavingEachIteration, withTumor, max_saves=5000):
+def _DEPIERRO_CPU(SMatrix, y, numIterations, beta, sigma, isSavingEachIteration, tumor_str, device, max_saves, show_logs=True):
     try:
-        if Omega != PotentialType.QUADRATIC:
-            raise ValueError("Depierro95 optimizer only supports QUADRATIC potential function.")
         if beta is None or sigma is None:
             raise ValueError("Depierro95 optimizer requires beta and sigma parameters.")
 
@@ -122,9 +156,10 @@ def _DEPIERRO_CPU(SMatrix, y, Omega, numIterations, beta, sigma, isSavingEachIte
             if save_indices[-1] != numIterations - 1:
                 save_indices.append(numIterations - 1)
 
-        description = f"AOT-BioMaps -- Bayesian Reconstruction Tomography: DE PIERRO (Sparse QUADRATIC β:{beta:.4f}, σ:{sigma:.4f}) ---- {'WITH' if withTumor else 'WITHOUT'} TUMOR ---- processing on single CPU ----"
+        description = f"AOT-BioMaps -- Bayesian Reconstruction Tomography: DE PIERRO (Sparse QUADRATIC β:{beta:.4f}, σ:{sigma:.4f}) ---- {tumor_str} TUMOR ---- processing on single CPU ----"
 
-        for p in trange(numIterations, desc=description):
+        iterator = trange(numIterations, desc=description) if show_logs else range(numIterations)
+        for it in iterator:
             theta_p = matrix_theta[-1]
             theta_p_flat = theta_p.reshape(-1)
             q_flat = np.dot(A_flat, theta_p_flat)
@@ -143,9 +178,9 @@ def _DEPIERRO_CPU(SMatrix, y, Omega, numIterations, beta, sigma, isSavingEachIte
             theta_p_plus_1_flat = np.clip(theta_p_plus_1_flat, a_min=0, a_max=None)
             theta_next = theta_p_plus_1_flat.reshape(Z, X)
             matrix_theta[-1] = theta_next
-            if isSavingEachIteration and p in save_indices:
+            if isSavingEachIteration and it in save_indices:
                 I_reconMatrix.append(theta_next.copy())
-                saved_indices.append(p)
+                saved_indices.append(it)
 
         if isSavingEachIteration:
             return I_reconMatrix, saved_indices

AOT_biomaps/AOT_Recon/AOT_Optimizers/LS.py

@@ -2,6 +2,7 @@ from AOT_biomaps.Config import config
 import torch
 import numpy as np
 from tqdm import trange
+from AOT_biomaps.AOT_Recon.ReconTools import calculate_memory_requirement, check_gpu_memory
 
 def LS(
     SMatrix,
@@ -12,6 +13,7 @@ def LS(
     withTumor=True,
     device=None,
     max_saves=5000,
+    show_logs=True
 ):
     """
     Least Squares reconstruction using Projected Gradient Descent (PGD) with non-negativity constraint.
@@ -20,15 +22,15 @@ def LS(
     tumor_str = "WITH" if withTumor else "WITHOUT"
     # Force GPU usage for now
     if device is None:
-        if not torch.cuda.is_available():
+        if torch.cuda.is_available() and check_gpu_memory(config.select_best_gpu(), calculate_memory_requirement(SMatrix, y), show_logs=show_logs):
             raise RuntimeError("CUDA is required for this implementation.")
         device = torch.device(f"cuda:{config.select_best_gpu()}")
     else:
         if device.type != "cuda":
             raise RuntimeError("Only GPU implementation is available for now.")
-    return _LS_GPU_stable(SMatrix, y, numIterations, alpha, isSavingEachIteration, tumor_str, max_saves)
+    return _LS_GPU_stable(SMatrix, y, numIterations, alpha, isSavingEachIteration, tumor_str, max_saves, show_logs=show_logs)
 
-def _LS_GPU_stable(SMatrix, y, numIterations, alpha, isSavingEachIteration, tumor_str, max_saves=5000):
+def _LS_GPU_stable(SMatrix, y, numIterations, alpha, isSavingEachIteration, tumor_str, max_saves=5000, show_logs=True):
     """
     Stable GPU implementation of LS using projected gradient descent with diagonal preconditioner.
     """
@@ -65,13 +67,14 @@ def _LS_GPU_stable(SMatrix, y, numIterations, alpha, isSavingEachIteration, tumo
     AT_r = torch.empty(ZX, device=device)
     description = f"AOT-BioMaps -- Stable LS Reconstruction ---- {tumor_str} TUMOR ---- GPU {torch.cuda.current_device()}"
 
-    for k in trange(numIterations, desc=description):
+    iterator = trange(numIterations, desc=description) if show_logs else range(numIterations)
+    for it in iterator:
         # Calcul du résidu (inplace)
         torch.matmul(A_flat, lambda_k, out=r_k)
         r_k = y_flat - r_k
-        if isSavingEachIteration and k in save_indices:
+        if isSavingEachIteration and it in save_indices:
             lambda_history.append(lambda_k.clone().reshape(Z, X) * (norm_y / norm_A))
-            saved_indices.append(k)
+            saved_indices.append(it)
 
         # Gradient préconditionné (inplace)
         torch.matmul(A_flat.T, r_k, out=AT_r)