Skip to content

Add U-Mamba architecture to MONAI #8433

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 4 commits into
base: dev
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions monai/networks/nets/__init__.py
Original file line number Diff line number Diff line change
Expand Up @@ -144,3 +144,4 @@
from .vnet import VNet
from .voxelmorph import VoxelMorph, VoxelMorphUNet
from .vqvae import VQVAE
from .u_mamba import UMambaUNet
110 changes: 110 additions & 0 deletions monai/networks/nets/u_mamba.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,110 @@
# Copyright (c) MONAI Consortium
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
# http://www.apache.org/licenses/LICENSE-2.0
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import torch
import torch.nn as nn
import torch.nn.functional as F

# Simple placeholder for the SSM (Mamba-like block)
class SSMBlock(nn.Module):
def __init__(self, dim):
super().__init__()
self.linear1 = nn.Linear(dim, dim)
self.linear2 = nn.Linear(dim, dim)

def forward(self, x):
# x: (B, L, C)
return self.linear2(torch.silu(self.linear1(x)))

class UMambaBlock(nn.Module):
def __init__(self, in_channels, hidden_channels):
super().__init__()
self.conv_res1 = nn.Sequential(
nn.Conv3d(in_channels, in_channels, kernel_size=3, padding=1),
nn.InstanceNorm3d(in_channels),
nn.LeakyReLU(),
)
self.conv_res2 = nn.Sequential(
nn.Conv3d(in_channels, in_channels, kernel_size=3, padding=1),
nn.InstanceNorm3d(in_channels),
nn.LeakyReLU(),
)

self.layernorm = nn.LayerNorm(hidden_channels)
self.linear1 = nn.Linear(in_channels, hidden_channels)
self.linear2 = nn.Linear(hidden_channels, in_channels)
self.conv1d = nn.Conv1d(hidden_channels, hidden_channels, kernel_size=3, padding=1)
self.ssm = SSMBlock(hidden_channels)

def forward(self, x):
# x: (B, C, H, W, D)
residual = x
x = self.conv_res1(x)
x = self.conv_res2(x) + residual

B, C, H, W, D = x.shape
x_flat = x.view(B, C, -1).permute(0, 2, 1) # (B, L, C)
x_norm = self.layernorm(x_flat)
x_proj = self.linear1(x_norm)

x_silu = torch.silu(x_proj)
x_ssm = self.ssm(x_silu)
x_conv1d = self.conv1d(x_proj.permute(0, 2, 1)).permute(0, 2, 1)

x_combined = torch.silu(x_conv1d) * torch.silu(x_ssm)
x_out = self.linear2(x_combined)
x_out = x_out.permute(0, 2, 1).view(B, C, H, W, D)

return x + x_out # Residual connection

class ResidualBlock(nn.Module):
def __init__(self, channels):
super().__init__()
self.block = nn.Sequential(
nn.Conv3d(channels, channels, kernel_size=3, padding=1),
nn.BatchNorm3d(channels),
nn.ReLU(),
nn.Conv3d(channels, channels, kernel_size=3, padding=1),
nn.BatchNorm3d(channels),
)

def forward(self, x):
return F.relu(x + self.block(x))

class UMambaUNet(nn.Module):
def __init__(self, in_channels=1, out_channels=1, base_channels=32):
super().__init__()
self.enc1 = UMambaBlock(in_channels, base_channels)
self.down1 = nn.Conv3d(base_channels, base_channels*2, kernel_size=3, stride=2, padding=1)

self.enc2 = UMambaBlock(base_channels*2, base_channels*2)
self.down2 = nn.Conv3d(base_channels*2, base_channels*4, kernel_size=3, stride=2, padding=1)

self.bottleneck = UMambaBlock(base_channels*4, base_channels*4)

self.up2 = nn.ConvTranspose3d(base_channels*4, base_channels*2, kernel_size=2, stride=2)
self.dec2 = ResidualBlock(base_channels*4)

self.up1 = nn.ConvTranspose3d(base_channels*2, base_channels, kernel_size=2, stride=2)
self.dec1 = ResidualBlock(base_channels*2)

self.final = nn.Conv3d(base_channels, out_channels, kernel_size=1)

def forward(self, x):
x1 = self.enc1(x)
x2 = self.enc2(self.down1(x1))
x3 = self.bottleneck(self.down2(x2))

x = self.up2(x3)
x = self.dec2(torch.cat([x, x2], dim=1))
x = self.up1(x)
x = self.dec1(torch.cat([x, x1], dim=1))
return self.final(x)
22 changes: 22 additions & 0 deletions tests/test_networks_u_mamba.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,22 @@
import unittest
import torch
from monai.networks.nets import UMambaUNet

class TestUMamba(unittest.TestCase):
def test_forward_shape(self):
# Set up input dimensions and model
input_tensor = torch.randn(2, 1, 16, 64, 64)
model = UMambaUNet(in_channels=1, out_channels=2)
output = model(input_tensor)
self.assertEqual(output.shape, (2, 2, 16, 64, 64))

def test_script(self):
# Test JIT scripting if supported
model = UMambaUNet(in_channels=1, out_channels=2)
scripted = torch.jit.script(model)
x = torch.randn(1, 1, 64, 64)
out = scripted(x)
self.assertEqual(out.shape, (1, 2, 64, 64))

if __name__ == "__main__":
unittest.main()
Loading