model.py

"""
VAE model definition.
Input: (B, 4, 256, 256) — 4 slices (3D CT label/mask).
Output: decode(z) -> (B, 4, 256, 256) — 4 reconstructed slices.
"""
import torch
import torch.nn as nn


class Conv(nn.Module):
    def __init__(self, in_channels: int, out_channels: int, kernel_size: int, stride: int = 1, padding: int = 0):
        super().__init__()
        self.conv = nn.Sequential(
            nn.Conv2d(in_channels, out_channels, kernel_size, stride, padding, bias=False),
            nn.BatchNorm2d(out_channels),
            nn.LeakyReLU(inplace=True),
        )

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        return self.conv(x)


class ConvTranspose(nn.Module):
    def __init__(self, in_channels: int, out_channels: int, kernel_size: int, stride: int = 1, padding: int = 0):
        super().__init__()
        self.conv = nn.Sequential(
            nn.ConvTranspose2d(in_channels, out_channels, kernel_size, stride, padding, bias=False),
            nn.BatchNorm2d(out_channels),
            nn.LeakyReLU(inplace=True),
        )

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        return self.conv(x)


class VAE(nn.Module):
    """VAE: 4-channel input (4 slices) -> latent -> 4-channel output (4 slices)."""

    def __init__(self, base: int = 64):
        super().__init__()
        self.base = base

        self.encoder = nn.Sequential(
            Conv(4, base, 3, stride=2, padding=1),
            Conv(base, 2 * base, 3, padding=1),
            Conv(2 * base, 2 * base, 3, stride=2, padding=1),
            Conv(2 * base, 2 * base, 3, padding=1),
            Conv(2 * base, 2 * base, 3, stride=2, padding=1),
            Conv(2 * base, 4 * base, 3, padding=1),
            Conv(4 * base, 4 * base, 3, stride=2, padding=1),
            Conv(4 * base, 4 * base, 3, padding=1),
            Conv(4 * base, 4 * base, 3, stride=2, padding=1),
            nn.Conv2d(4 * base, 64 * base, 8),
            nn.LeakyReLU(inplace=True),
        )

        self.encoder_mu = nn.Conv2d(64 * base, 32 * base, 1)
        self.encoder_logvar = nn.Conv2d(64 * base, 32 * base, 1)

        self.decoder = nn.Sequential(
            nn.Conv2d(32 * base, 64 * base, 1),
            ConvTranspose(64 * base, 4 * base, 8),
            Conv(4 * base, 4 * base, 3, padding=1),
            ConvTranspose(4 * base, 4 * base, 4, stride=2, padding=1),
            Conv(4 * base, 4 * base, 3, padding=1),
            ConvTranspose(4 * base, 4 * base, 4, stride=2, padding=1),
            Conv(4 * base, 2 * base, 3, padding=1),
            ConvTranspose(2 * base, 2 * base, 4, stride=2, padding=1),
            Conv(2 * base, 2 * base, 3, padding=1),
            ConvTranspose(2 * base, 2 * base, 4, stride=2, padding=1),
            Conv(2 * base, base, 3, padding=1),
            ConvTranspose(base, base, 4, stride=2, padding=1),
            nn.Conv2d(base, 4, 3, padding=1),
            nn.Sigmoid(),
        )

    def encode(self, x: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor]:
        x = self.encoder(x)
        return self.encoder_mu(x), self.encoder_logvar(x)

    def reparameterize(self, mu: torch.Tensor, logvar: torch.Tensor) -> torch.Tensor:
        """Standard VAE reparameterization: z = mu + std * eps. For inference (eval mode), return mu for deterministic output."""
        if not self.training:
            return mu
        std = torch.exp(0.5 * logvar)
        eps = torch.randn_like(std)
        return mu + std * eps

    def decode(self, z: torch.Tensor) -> torch.Tensor:
        return self.decoder(z)

    def forward(self, x: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
        mu, logvar = self.encode(x)
        z = self.reparameterize(mu, logvar)
        return self.decode(z), mu, logvar