Pretrained Neural Network Models¶
Despite the source code of TIAToolbox being held under a permissive license, the licenses of model weights are dependent on the datasets that they are trained on. We provide the licenses associated with the utilised datasets, but recommend that users also do their own due diligence for confirmation.
Patch Classification¶
Kather Patch Dataset¶
The following models are trained using Kather Dataset.
Model weights obtained from training on the Kather100K dataset are held under the Creative Commons Attribution 4.0 International License.
They share the same input output configuration defined below:
Input Output Configuration Details
from tiatoolbox.models import IOPatchPredictorConfig
ioconfig = IOPatchPredictorConfig(
patch_input_shape=(224, 224),
stride_shape=(224, 224),
input_resolutions=[{"resolution": 0.5, "units": "mpp"}]
)
Model names
alexnet-kather100k
resnet18-kather100k
resnet34-kather100k
resnet50-kather100k
resnet101-kather100k
resnext50_32x4d-kather100k
resnext101_32x8d-kather100k
wide_resnet50_2-kather100k
wide_resnet101_2-kather100k
densenet121-kather100k
densenet161-kather100k
densenet169-kather100k
densenet201-kather100k
mobilenet_v2-kather100k
mobilenet_v3_large-kather100k
mobilenet_v3_small-kather100k
googlenet-kather100k
Patch Camelyon (PCam) Dataset¶
The following models are trained using the PCam dataset. The model weights obtained from training on the PCam dataset are held under the CC0 License. They share the same input output configuration defined below:
Input Output Configuration Details
from tiatoolbox.models import IOPatchPredictorConfig
ioconfig = IOPatchPredictorConfig(
patch_input_shape=(96, 96),
stride_shape=(96, 96),
input_resolutions=[{"resolution": 1.0, "units": "mpp"}]
)
Model names
alexnet-pcam
resnet18-pcam
resnet34-pcam
resnet50-pcam
resnet101-pcam
resnext50-pcam
resnext101-pcam
wide_resnet50_2-pcam
wide_resnet101_2-pcam
densenet121-pcam
densenet161-pcam
densenet169-pcam
densenet201-pcam
mobilenet_v2-pcam
mobilenet_v3_large-pcam
mobilenet_v3_small-pcam
googlenet-pcam
Semantic Segmentation¶
Tissue Masking¶
The following models are trained using internal data of TIA Centre and are held under the Creative Commons Attribution-NonCommercial-ShareAlike Version 4 (CC BY-NC-SA 4.0) License. They share the same input output configuration defined below:
Input Output Configuration Details
from tiatoolbox.models import IOSegmentorConfig
ioconfig = IOSegmentorConfig(
input_resolutions=[
{'units': 'mpp', 'resolution': 2.0}
],
output_resolutions=[
{'units': 'mpp', 'resolution': 2.0}
],
patch_input_shape=(1024, 1024),
patch_output_shape=(512, 512),
stride_shape=(256, 256),
save_resolution={'units': 'mpp', 'resolution': 8.0}
)
Model names
fcn-tissue_mask
Breast Cancer¶
The following models are trained using the BCSS dataset. The model weights obtained from training on the BCSS dataset are held under the CC0 License. They share the same input output configuration defined below:
Input Output Configuration Details
from tiatoolbox.models import IOSegmentorConfig
ioconfig = IOSegmentorConfig(
input_resolutions=[
{'units': 'mpp', 'resolution': 0.25}
],
output_resolutions=[
{'units': 'mpp', 'resolution': 0.25}
],
patch_input_shape=(1024, 1024),
patch_output_shape=(512, 512),
stride_shape=(256, 256),
save_resolution={'units': 'mpp', 'resolution': 0.25}
)
Model names
fcn_resnet50_unet-bcss
Nucleus Instance Segmentation¶
PanNuke Dataset¶
We provide the following models trained using the PanNuke dataset. All model weights trained on PanNuke are held under the Creative Commons Attribution-NonCommercial-ShareAlike Version 4 (CC BY-NC-SA 4.0) License. The input output configuration is as follows:
Input Output Configuration Details
from tiatoolbox.models import IOSegmentorConfig
ioconfig = IOSegmentorConfig(
input_resolutions=[
{'units': 'mpp', 'resolution': 0.25}
],
output_resolutions=[
{'units': 'mpp', 'resolution': 0.25},
{'units': 'mpp', 'resolution': 0.25},
{'units': 'mpp', 'resolution': 0.25}
],
margin=128
tile_shape=[1024, 1024]
patch_input_shape=(256, 256),
patch_output_shape=(164, 164),
stride_shape=(164, 164),
save_resolution={'units': 'mpp', 'resolution': 0.25}
)
Model names
hovernet_fast-pannuke
Output Nuclear Classes
0: Background
1: Neoplastic
2: Inflammatory
3: Connective
4: Dead
5: Non-Neoplastic Epithelial
MoNuSAC Dataset¶
We provide the following models trained using the MoNuSAC dataset. All model weights trained on MoNuSAC are held under the Creative Commons Attribution-NonCommercial-ShareAlike Version 4 (CC BY-NC-SA 4.0) License. The input output configuration is as follows:
Input Output Configuration Details
from tiatoolbox.models import IOSegmentorConfig
ioconfig = IOSegmentorConfig(
input_resolutions=[
{'units': 'mpp', 'resolution': 0.25}
],
output_resolutions=[
{'units': 'mpp', 'resolution': 0.25},
{'units': 'mpp', 'resolution': 0.25},
{'units': 'mpp', 'resolution': 0.25}
],
margin=128
tile_shape=[1024, 1024]
patch_input_shape=(256, 256),
patch_output_shape=(164, 164),
stride_shape=(164, 164),
save_resolution={'units': 'mpp', 'resolution': 0.25}
)
Model names
hovernet_fast-monusac
Output Nuclear Classes
0: Background
1: Epithelial
2: Lymphocyte
3: Macrophage
4: Neutrophil
CoNSeP Dataset¶
We provide the following models trained using the CoNSeP dataset. The model weights obtained from training on the CoNSeP dataset are held under the Apache 2.0 License. The input output configuration is as follows:
Input Output Configuration Details
from tiatoolbox.models import IOSegmentorConfig
ioconfig = IOSegmentorConfig(
input_resolutions=[
{'units': 'mpp', 'resolution': 0.25}
],
output_resolutions=[
{'units': 'mpp', 'resolution': 0.25},
{'units': 'mpp', 'resolution': 0.25},
{'units': 'mpp', 'resolution': 0.25}
],
margin=128
tile_shape=[1024, 1024]
patch_input_shape=(270, 270),
patch_output_shape=(80, 80),
stride_shape=(80, 80),
save_resolution={'units': 'mpp', 'resolution': 0.25}
)
Model names
hovernet_original-consep
Output Nuclear Classes
0: Background
1: Epithelial
2: Inflammatory
3: Spindle-Shaped
4: Miscellaneous
Input Output Configuration Details
from tiatoolbox.models import IOSegmentorConfig
ioconfig = IOSegmentorConfig(
input_resolutions=[
{'units': 'mpp', 'resolution': 0.25}
],
output_resolutions=[
{'units': 'mpp', 'resolution': 0.25}
],
tile_shape=[2048, 2048]
patch_input_shape=(252, 252),
patch_output_shape=(252, 252),
stride_shape=(150, 150),
save_resolution={'units': 'mpp', 'resolution': 0.25}
)
Model names
micronet_hovernet-consep
Kumar Dataset¶
We provide the following models trained using the Kumar dataset. All model weights trained on Kumar are held under the Creative Commons Attribution-NonCommercial-ShareAlike Version 4 (CC BY-NC-SA 4.0) License. The Kumar dataset does not contain nuclear class information, and so TIAToolbox pretrained models based on Kumar for nuclear segmentation, will only perform segmentation and not classification. The input output configuration is as follows:
Input Output Configuration Details
from tiatoolbox.models import IOSegmentorConfig
ioconfig = IOSegmentorConfig(
input_resolutions=[
{'units': 'mpp', 'resolution': 0.25}
],
output_resolutions=[
{'units': 'mpp', 'resolution': 0.25},
{'units': 'mpp', 'resolution': 0.25},
{'units': 'mpp', 'resolution': 0.25}
],
margin=128
tile_shape=[1024, 1024]
patch_input_shape=(270, 270),
patch_output_shape=(80, 80),
stride_shape=(80, 80),
save_resolution={'units': 'mpp', 'resolution': 0.25}
)
Model names
hovernet_original_kumar
Nucleus Detection¶
CRCHisto Dataset¶
We provide the following models trained using the CRCHisto dataset. All model weights trained on CRCHisto are held under the Creative Commons Attribution-NonCommercial-ShareAlike Version 4 (CC BY-NC-SA 4.0) License. The input output configuration is as follows:
Input Output Configuration Details
from tiatoolbox.models import IOPatchPredictorConfig
ioconfig = IOPatchPredictorConfig(
patch_input_shape=(31, 31),
stride_shape=(8, 8),
input_resolutions=[{"resolution": 0.25, "units": "mpp"}]
)
Model names
sccnn-crchisto
Input Output Configuration Details
from tiatoolbox.models import IOPatchPredictorConfig
ioconfig = IOPatchPredictorConfig(
patch_input_shape=(252, 252),
stride_shape=(150, 150),
input_resolutions=[{"resolution": 0.25, "units": "mpp"}]
)
Model names
mapde-crchisto
CoNIC Dataset¶
We provide the following models trained using the CoNIC dataset. All model weights trained on CoNIC are held under the Creative Commons Attribution-NonCommercial-ShareAlike Version 4 (CC BY-NC-SA 4.0) License. The input output configuration is as follows:
Input Output Configuration Details
from tiatoolbox.models import IOPatchPredictorConfig
ioconfig = IOPatchPredictorConfig(
patch_input_shape=(31, 31),
stride_shape=(8, 8),
input_resolutions=[{"resolution": 0.25, "units": "mpp"}]
)
Model names
sccnn-conic
Input Output Configuration Details
from tiatoolbox.models import IOPatchPredictorConfig
ioconfig = IOPatchPredictorConfig(
patch_input_shape=(252, 252),
stride_shape=(150, 150),
input_resolutions=[{"resolution": 0.25, "units": "mpp"}]
)
Model names
mapde-conic
Multi-Task Segmentation¶
Oral Epithelial Dysplasia (OED) Dataset¶
We provide the following model trained using a private OED dataset. The model outputs nuclear instance segmentation and classification results, as well as semantic segmentation of epithelial layers. All model weights trained on the private OED dataset are held under the Creative Commons Attribution-NonCommercial-ShareAlike Version 4 (CC BY-NC-SA 4.0) License. The model uses the following input output configuration:
Input Output Configuration Details
from tiatoolbox.models import IOSegmentorConfig
ioconfig = IOSegmentorConfig(
input_resolutions=[
{'units': 'mpp', 'resolution': 0.5}
],
output_resolutions=[
{'units': 'mpp', 'resolution': 0.5},
{'units': 'mpp', 'resolution': 0.5},
{'units': 'mpp', 'resolution': 0.5},
{'units': 'mpp', 'resolution': 0.5}
],
margin=128
tile_shape=[1024, 1024]
patch_input_shape=(256, 256),
patch_output_shape=(164, 164),
stride_shape=(164, 164),
save_resolution={'units': 'mpp', 'resolution': 0.5}
)
Model names
hovernetplus-oed
Output Nuclear Classes
0: Background
1: Other
2: Epithelial
Output Region Classes
0: Background
1: Other Tissue
2: Basal Epithelium
3: (Core) Epithelium
4: Keratin