Unified deep learning model for color unmixing

Abstract

The present disclosure relates to stain unmixing of a multiplex digital pathology image into single stain images by processing the multiplex image with a unified deep learning model. The unified deep learning model comprises of a generator and a discriminator. The prior knowledge on the imaging physics or stain chemistry may be incorporated by adding constraints in the model. The generator may be trained and configured to generate a stain contribution map corresponding to each stain in the multiplex image. The discriminator may be configured to discriminate real or synthetic images and to classify stain class of input single stain image. The model may be trained in an unsupervised setting using a dataset of unpaired multiplex and single stain images. During the training, both the generator and the discriminator learns adversarially based on one or more loss functions. The stain contribution maps produced by the generator can be multiplied with the multiplex image to obtain single stain images.

Description

[0001]- - ,, .. . , , , - . .., , ,. . ., , , . , ., -, , , .,. - , .., .. , . ,- - . - . , - , , . - , . , , , .. , . - .. . . , , .. .. , , . - . . , - .. ,- , . - ., ,, ,, . , , . . ,- , . . ,. . , . ,,, , , - . , , - , , , - . . -, , ,, . , . . . . - , . . , , - , . , , - , , , . - ,- - . , . , ,. , , , . . . . . . - . .. ... ... - . . . . .. . . . . , . . , . - - . , “ . ” “ ” “ ” . , “ ” . , “ ” . “ ” - , , ., , “ ”, , , , .., . . , “” , . . . , ., , “ ” . , . , “,” , . . , – , – - , . . , - . . - , . .. -- . , , .., , , , . . , . ,, , . , . , . . ,. , , . . ., , ,.. . ,, , , . , . , .., - , , . . , , , . , . . , , - -- ., , , , , . , . . ..,, , . , . .-. .., - , . -^^^",1. -^^^., .. , - , , , - , ^^'. .. , ^^*., ^^*^^'. - , ,- . ,, . , - .., –, .., . , - ..., . , , . , . . - . ,, .. , - - , ,, , ,,. .., , , . . . ,- , , . , - . . , . - . , ,- . ., , , , . , , “ . . ” , .., , , . . , , . . - - . . , . , , , . , . ™, ™-, - . , -- , . , , , , - , ,, . , , . , . . , . . , ., . .. - . . , - , , .. , , , . , , ..,, . . . . . . . . . , , , . . , . . ,- , - . , . - , ...,- , , ..,, , . , , , , , , , ., , . , .- , , , , , . .., . , , - . , , , , - , , , . , , - , .. . - . . , , , , , . ,, , . , , , . , , . . . . , , , , ...,., , , ..., . .., .., . .., ,- . , , . . , ..., . , - , . - . ..,, , .., - . , ,. -, , , . .., , ,. , , .., . .. . .., ' , - , , , .., , .., ., - - . - " ", , . .., . , , . , .., . , . - -. , . .., . - . - ,- , , . , , , .., ,,- ,.... - - , . “”.., , .., . , . ..,‘’ , “” ,- . - , ,., . , , ., , , . . , . . . . .. - . ., , - , , . . . , – , ^^ ^^+" = − ^^ = ^.^^* ^^^*=^.^^*^^^^^^^^^ ^ , * * ,. , , . , . , , . , . .^^ ^^+" ^→ ^=^*- ^=^*- ^^+". - - .. - ^^^^, ^=^ → ^^^ , ^, **,"^.-. , . , . , - . - , , , . ,- .., . ^^=^* =^^^* ^ ^^+"^^^*, ^ ^^ ^,^^^^^^^^^^^^ ^.^^ - +" . ,..,,,^^^,,-^^ ^^^^^^^^^^^^ → ^.*^^*^^,, ^^ =,^^^ ^ . .. -^=^* -,^^*., . -..., ^^ - - ,. ^^" ^^".^^, ^^, ^^+" , ^^ = ^^ ^^^ ^^'^^ ^^^^"^^^^^"^^, ^^^'!"" "^ ^^^^^^^, ^^ ^- ,1, - -, . , , -. ^^*^^, ^^, ^^+" ,,^^ = ^^ ^^^ ^^^^^^,^^'^ ^'^^'!"^ ^^^""^^"^ ^^^^^ ^ ^^^, ^^*.. , ., * ^^'. , ,. , . - ^̂^*^^ ^^' =^ ^ ^̂^*^^ − ^^*^^ ^,^^^^^^^^^ ^^ = ^^^ ^^ ^, ^=^*^^, ^-^ -, ,*., , . , , - .., – - .., . ,- ..., . , ., . , ..., . , . , , , .., . , . , . .., . , . - , . , .., - . , , -, . , . - .. , - , , , ,. .., . , . -- , - , - , -, .. . - . . , - , , . - .., . .., , , - .., . ., - -. .- , , . - . .., . . ., . . ^ .., .. , , . - . . ., , . , , . .^ , . . . . , , . , - ,, - , . ,- , , . . , , , . , . .^ . . . , , . .^ , ^ - . . . , . , , , . , . . . , . - - . ,, . - .^ ,- .. ,, , . - , . , ...^, - , . , ,, - . , --, . , . , - ,.. ^ . , .., . , , -- ., . .^ . - . . , ,- ., - . . . -- , , _- . ..,-.- . - --- . . . .. , , . , , ,- , - - . . - . - .. , - . . . . - , - ,- - . - . . . , . , , . - - , . . .-., - . - , , -., . . -- . . , , -, - . , - - . - , ,. , , ,, . , . ,, . . .. , , , , , , , , . ,- , , .

Claims

- . - , , , , , . . - , - . . - , , -., , ,- . - ,, . . - , . . - , . . - . , . - , ,- . . -, , , . . - , , . , ,- . , - , , , , , , . . , -. . . ,- ,. , , . - . -, - , , , , , . . - , - . . - ,- , -., , , . - , , . . -, ,- . . ,- , , .

Images