[BUG] Can not use kspaceFirstOrder-CUDA #622
Description
Describe the bug
Hi, I am new to this project and I encounter some problems when I try to use the kspaceFirstOrder-CUDA. Here are my codes:
import rospy
from sensor_msgs.msg import Image
from cv_bridge import CvBridge
import numpy as np
from kwave import kWaveGrid, kSensor, kSource, medium as kw_medium, kspaceFirstOrder3D
class PhotoacousticReconstruction:
def __init__(self):
self.bridge = CvBridge()
self.sub = rospy.Subscriber('/slice_image', Image, self.callback)
self.pub = rospy.Publisher('/reconstructed_image', Image, queue_size=1)
# Define k-Wave grid
dx = 0.2e-3 # [m]
Nx, Ny, Nz = 250, 200, 200
self.kgrid = kWaveGrid(Nx=Nx, dx=dx, Ny=Ny, dy=dx, Nz=Nz, dz=dx)
# Define medium
self.medium = kw_medium(sound_speed=1540, density=1000, alpha_coeff=0.75, alpha_power=1.5)
# Define sensor (placeholder for sensor mask)
element_num = 128
element_pitch = 0.3e-3 # [m]
self.sensor = kSensor()
self.sensor.mask = np.zeros((Nx, Ny, Nz), dtype=bool)
# Define sensor mask based on transducer array
for ind in range(element_num):
x_pos = -(element_num * element_pitch / 2 - element_pitch / 2) + (ind * element_pitch)
y_idx = int((x_pos + (Ny * dx) / 2) / dx) # Center grid in y
if 0 <= y_idx < Ny:
self.sensor.mask[0, y_idx, 0] = True # Approximate each transducer as a point
self.sensor.record = {'p': True} # Record pressure during time reversal
# Define source (zero for time reversal)
self.source = kSource()
self.source.p0 = np.zeros((Nx, Ny, Nz))
# Define time array
self.kgrid.t_array = self.kgrid.makeTime(self.medium.sound_speed)
def callback(self, data):
try:
# Convert ROS Image to NumPy array
cv_image = self.bridge.imgmsg_to_cv2(data, "32FC1") # Assuming 32-bit float
num_sensor_points, Nt = cv_image.shape
# Verify sensor data matches mask
if num_sensor_points != np.sum(self.sensor.mask):
rospy.logwarn("Sensor data size does not match sensor mask points")
return
# Set time reversal boundary data
self.sensor.time_reversal_boundary_data = cv_image
# Run time reversal simulation
p0_recon = kspaceFirstOrder3D(self.kgrid, self.medium, self.source, self.sensor)
# Extract reconstructed initial pressure
reconstructed_p0 = p0_recon.p
# Extract 2D slice at z = Nz/2
slice_image = reconstructed_p0[:, :, self.kgrid.Nz // 2]
# Publish reconstructed image
slice_image_msg = self.bridge.cv2_to_imgmsg(slice_image.astype(np.float32), "32FC1")
self.pub.publish(slice_image_msg)
except Exception as e:
rospy.logerr(f"Error processing image: {str(e)}")
def run(self):
rospy.spin()
if __name__ == '__main__':
try:
rospy.init_node('photoacoustic_reconstruction', anonymous=True)
node = PhotoacousticReconstruction()
node.run()
except rospy.ROSInterruptException:
pass
And here are the errors I encounter:
home/thera/anaconda3/envs/kwave/lib/python3.10/site-packages/skimage/transform/_warps.py:2: UserWarning: A NumPy version >=1.23.5 and <2.5.0 is required for this version of SciPy (detected version 1.23.0)
from scipy import ndimage as ndi
[INFO] [1752999585.895047]: sensor_mask shape: (250, 200, 200), non-zero points: 7000
[WARN] [1752999585.901278]: sensor_mask has 7000 non-zero points, expected 128. Adjusting mask...
[INFO] [1752999585.908868]: Adjusted sensor_mask non-zero points: 128
┌───────────────────────────────────────────────────────────────┐
│ kspaceFirstOrder-CUDA v1.3 │
├───────────────────────────────────────────────────────────────┤
│ Reading simulation configuration: Done │
│ Selected GPU device id: Failed │
└───────────────────────────────────────────────────────────────┘
┌───────────────────────────────────────────────────────────────┐
│ !!! K-Wave experienced a fatal error !!! │
├───────────────────────────────────────────────────────────────┤
│ Error: All CUDA-capable devices are busy or unavailable. │
├───────────────────────────────────────────────────────────────┤
│ Execution terminated │
└───────────────────────────────────────────────────────────────┘
[ERROR] [1752999588.451475]: bad callback: <function main.<locals>.image_callback at 0x7fcc170c8820>
Traceback (most recent call last):
File "/opt/ros/noetic/lib/python3/dist-packages/rospy/topics.py", line 750, in _invoke_callback
cb(msg)
File "/home/thera/expert_simulator_pa_7_5/src/ct_to_us_simulator/scripts/kwave_ros_simulator.py", line 96, in image_callback
sensor_data = kspaceFirstOrder3DG(kgrid, source, sensor, medium, simulation_options, execution_options)
File "/home/thera/Downloads/k-wave-python-master/kwave/kspaceFirstOrder3D.py", line 70, in kspaceFirstOrder3DG
sensor_data = kspaceFirstOrder3D(
File "/home/thera/Downloads/k-wave-python-master/kwave/kspaceFirstOrder3D.py", line 375, in kspaceFirstOrder3D
sensor_data = executor.run_simulation(k_sim.options.input_filename, k_sim.options.output_filename, options=executor_options)
File "/home/thera/Downloads/k-wave-python-master/kwave/executor.py", line 53, in run_simulation
raise subprocess.CalledProcessError(proc.returncode, command, stdout, stderr)
subprocess.CalledProcessError: Command '['/home/thera/Downloads/k-wave-python-master/kwave/bin/linux/kspaceFirstOrder-CUDA', '-i', '/tmp/20-Jul-2025-16-19-45_kwave_input.h5', '-o', '/tmp/20-Jul-2025-16-19-45_kwave_output.h5', '-t', '20', '--p_raw', '-s', '1']' returned non-zero exit status 1.
To Reproduce
Steps to reproduce the behavior:
just run the codes and reproduce the behavior.
Expected behavior
The output should generate a generated image.
Screenshots
If applicable, add screenshots to help explain your problem.
Desktop (please complete the following information):
- OS: Linux
- Version 20
- Python 3.10.0
- numpy 1.23.0
Thank you for your help