Class HoVerNet

monai/networks/nets/hovernet.py:412–609

HoVerNet model References: Graham, Simon et al. Hover-net: Simultaneous segmentation and classification of nuclei in multi-tissue histology images, Medical Image Analysis 2019 https://github.com/vqdang/hover_net https://pytorch.org/vision/main/models/generated/tor

class HoVerNet(nn.Module):
    """HoVerNet model

    References:
      Graham, Simon et al. Hover-net: Simultaneous segmentation
      and classification of nuclei in multi-tissue histology images,
      Medical Image Analysis 2019

      https://github.com/vqdang/hover_net
      https://pytorch.org/vision/main/models/generated/torchvision.models.resnet50.html

    This network is non-deterministic since it uses `torch.nn.Upsample` with ``UpsampleMode.NONTRAINABLE`` mode which
    is implemented with torch.nn.functional.interpolate(). Please check the link below for more details:
    https://pytorch.org/docs/stable/generated/torch.use_deterministic_algorithms.html#torch.use_deterministic_algorithms

    Args:
        mode: use original implementation (`HoVerNetMODE.ORIGINAL` or "original") or
          a faster implementation (`HoVerNetMODE.FAST` or "fast"). Defaults to `HoVerNetMODE.FAST`.
        in_channels: number of the input channel.
        np_out_channels: number of the output channel of the nucleus prediction branch.
        out_classes: number of the nuclear type classes.
        act: activation type and arguments. Defaults to relu.
        norm: feature normalization type and arguments. Defaults to batch norm.
        decoder_padding: whether to do padding on convolution layers in the decoders. In the conic branch
            of the referred repository, the architecture is changed to do padding on convolution layers in order to
            get the same output size as the input, and this changed version is used on CoNIC challenge.
            Please note that to get consistent output size, `HoVerNetMode.FAST` mode should be employed.
        dropout_prob: dropout rate after each dense layer.
        pretrained_url: if specifying, will loaded the pretrained weights downloaded from the url.
            There are two supported forms of weights:
            1. preact-resnet50 weights coming from the referred hover_net
            repository, each user is responsible for checking the content of model/datasets and the applicable licenses
            and determining if suitable for the intended use. please check the following link for more details:
            https://github.com/vqdang/hover_net#data-format
            2. standard resnet50 weights of torchvision. Please check the following link for more details:
            https://pytorch.org/vision/main/_modules/torchvision/models/resnet.html#ResNet50_Weights
        adapt_standard_resnet: if the pretrained weights of the encoder follow the original format (preact-resnet50), this
            value should be `False`. If using the pretrained weights that follow torchvision's standard resnet50 format,
            this value should be `True`.
        pretrained_state_dict_key: this arg is used when `pretrained_url` is provided and `adapt_standard_resnet` is True.
            It is used to extract the expected state dict.
        freeze_encoder: whether to freeze the encoder of the network.
    """

    Mode = HoVerNetMode
    Branch = HoVerNetBranch

    def __init__(
        self,
        mode: HoVerNetMode | str = HoVerNetMode.FAST,
        in_channels: int = 3,
        np_out_channels: int = 2,
        out_classes: int = 0,
        act: str | tuple = ("relu", {"inplace": True}),
        norm: str | tuple = "batch",
        decoder_padding: bool = False,
        dropout_prob: float = 0.0,
        pretrained_url: str | None = None,