IDaRS¶

tiatoolbox . models . architecture . idars . IDaRS

class IDaRS(backbone, num_classes=1)[source]¶

Initialise IDaRS and add custom preprocessing as used in the original paper [1].

The tiatoolbox model should produce the following results:

IDaRS performance measured by AUROC.¶
	MSI	TP53	BRAF	CIMP	CIN	HM
Bilal et al.	0.828	0.755	0.813	0.853	0.860	0.846
TIAToolbox	0.870	0.747	0.750	0.748	0.810	0.790

Parameters:

backbone (str) – Model name.
num_classes (int) – Number of classes output by model.

References

[1] Bilal, Mohsin, et al. “Development and validation of a weakly supervised deep learning framework to predict the status of molecular pathways and key mutations in colorectal cancer from routine histology images: a retrospective study.” The Lancet Digital Health 3.12 (2021): e763-e772.

Initialize IDaRS.

Methods

preproc

Define preprocessing steps.

Attributes

training

static preproc(image)[source]¶

Define preprocessing steps.

Parameters:: image (numpy.ndarray) – An image of shape HWC.
Returns:: An image of shape HWC.
Return type:: image (torch.Tensor)