Normalization of HE-stained histological images using cycle consistent generative adversarial networks
Background: Histological images show strong variance (e.g. illumination, color, staining quality) due to differences in image acquisition, tissue processing, staining, etc. This can impede downstream image analysis such as staining intensity evaluation or classification. Methods to reduce these vari...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article (Journal) |
| Language: | English |
| Published: |
06 August 2021
|
| In: |
Diagnostic pathology
Year: 2021, Volume: 16, Pages: 1-10 |
| ISSN: | 1746-1596 |
| DOI: | 10.1186/s13000-021-01126-y |
| Online Access: | Resolving-System, kostenfrei, Volltext: https://doi.org/10.1186/s13000-021-01126-y Verlag, kostenfrei, Volltext: https://diagnosticpathology.biomedcentral.com/articles/10.1186/s13000-021-01126-y 
|
| Author Notes: | Marlen Runz, Daniel Rusche, Stefan Schmidt, Martin R. Weihrauch, Jürgen Hesser and Cleo-Aron Weis |
MARC
| LEADER | 00000caa a2200000 c 4500 | ||
|---|---|---|---|
| 001 | 1848823711 | ||
| 003 | DE-627 | ||
| 005 | 20230801115720.0 | ||
| 007 | cr uuu---uuuuu | ||
| 008 | 230612s2021 xx |||||o 00| ||eng c | ||
| 024 | 7 | |a 10.1186/s13000-021-01126-y |2 doi | |
| 035 | |a (DE-627)1848823711 | ||
| 035 | |a (DE-599)KXP1848823711 | ||
| 035 | |a (OCoLC)1389537787 | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rda | ||
| 041 | |a eng | ||
| 084 | |a 33 |2 sdnb | ||
| 100 | 1 | |a Runz, Marlen |e VerfasserIn |0 (DE-588)1237754003 |0 (DE-627)176470939X |4 aut | |
| 245 | 1 | 0 | |a Normalization of HE-stained histological images using cycle consistent generative adversarial networks |c Marlen Runz, Daniel Rusche, Stefan Schmidt, Martin R. Weihrauch, Jürgen Hesser and Cleo-Aron Weis |
| 264 | 1 | |c 06 August 2021 | |
| 300 | |a 10 | ||
| 336 | |a Text |b txt |2 rdacontent | ||
| 337 | |a Computermedien |b c |2 rdamedia | ||
| 338 | |a Online-Ressource |b cr |2 rdacarrier | ||
| 500 | |a Gesehen am 12.06.2023 | ||
| 520 | |a Background: Histological images show strong variance (e.g. illumination, color, staining quality) due to differences in image acquisition, tissue processing, staining, etc. This can impede downstream image analysis such as staining intensity evaluation or classification. Methods to reduce these variances are called image normalization techniques. - Methods: In this paper, we investigate the potential of CycleGAN (cycle consistent Generative Adversarial Network) for color normalization in hematoxylin-eosin stained histological images using daily clinical data with consideration of the variability of internal staining protocol variations. The network consists of a generator network GB that learns to map an image X from a source domain A to a target domain B, i.e. GB : XA → XB. In addition, a discriminator network DB is trained to distinguish whether an image from domain B is real or generated. The same process is applied to another generator-discriminator pair (GA, DA), for the inverse mapping GA : XB → XA. Cycle consistency ensures that a generated image is close to its original when being mapped backwards (GA(GB(XA)) ≈ XA and vice versa). We validate the CycleGAN approach on a breast cancer challenge and a follicular thyroid carcinoma data set for various stain variations. We evaluate the quality of the generated images compared to the original images using similarity measures. In addition, we apply stain normalization on pathological lymph node data from our institute and test the gain from normalization on a ResNet classifier pre-trained on the Camelyon16 data set. - Results: Qualitative results of the images generated by our network are compared to original color distributions. Our evaluation indicates that by mapping images to a target domain, the similarity training images from that domain improves up to 96%. We also achieve a high cycle consistency for the generator networks by obtaining similarity indices greater than 0.9. When applying the CycleGAN normalization to HE-stain images from our institute the kappa-value of the ResNet-model that is only trained on Camelyon16 data is increased more than 50%. - Conclusions: CycleGANs have proven to efficiently normalize HE-stained images. The approach compensates for deviations resulting from image acquisition (e.g. different scanning devices) as well as from tissue staining (e.g. different staining protocols), and thus overcomes the staining variations in images from various institutions. (Continued on next page) | ||
| 700 | 1 | |a Rusche, Daniel |d 1994- |e VerfasserIn |0 (DE-588)1297866509 |0 (DE-627)1854032895 |4 aut | |
| 700 | 1 | |a Schmidt, Stefan |e VerfasserIn |4 aut | |
| 700 | 1 | |a Weihrauch, Martin |d 1971- |e VerfasserIn |0 (DE-588)121136752 |0 (DE-627)081110537 |0 (DE-576)29255477X |4 aut | |
| 700 | 1 | |a Hesser, Jürgen |d 1964- |e VerfasserIn |0 (DE-588)1020647353 |0 (DE-627)691291071 |0 (DE-576)361513739 |4 aut | |
| 700 | 1 | |a Weis, Cleo-Aron Thias |d 1985- |e VerfasserIn |0 (DE-588)1062803752 |0 (DE-627)806961198 |0 (DE-576)42022730X |4 aut | |
| 773 | 0 | 8 | |i Enthalten in |t Diagnostic pathology |d [Erscheinungsort nicht ermittelbar] : BioMed Central, 2006 |g 16(2021) vom: Aug., Artikel-ID 71, Seite 1-10 |h Online-Ressource |w (DE-627)503328960 |w (DE-600)2210518-9 |w (DE-576)260587265 |x 1746-1596 |7 nnas |a Normalization of HE-stained histological images using cycle consistent generative adversarial networks |
| 773 | 1 | 8 | |g volume:16 |g year:2021 |g month:08 |g elocationid:71 |g pages:1-10 |g extent:10 |a Normalization of HE-stained histological images using cycle consistent generative adversarial networks |
| 856 | 4 | 0 | |u https://doi.org/10.1186/s13000-021-01126-y |x Resolving-System |x Verlag |z kostenfrei |3 Volltext |
| 856 | 4 | 0 | |u https://diagnosticpathology.biomedcentral.com/articles/10.1186/s13000-021-01126-y |x Verlag |z kostenfrei |3 Volltext |
| 951 | |a AR | ||
| 992 | |a 20230612 | ||
| 993 | |a Article | ||
| 994 | |a 2021 | ||
| 998 | |g 1062803752 |a Weis, Cleo-Aron Thias |m 1062803752:Weis, Cleo-Aron Thias |d 60000 |e 60000PW1062803752 |k 0/60000/ |p 6 |y j | ||
| 998 | |g 1020647353 |a Hesser, Jürgen |m 1020647353:Hesser, Jürgen |d 60000 |d 63000 |e 60000PH1020647353 |e 63000PH1020647353 |k 0/60000/ |k 1/60000/63000/ |p 5 | ||
| 998 | |g 1297866509 |a Rusche, Daniel |m 1297866509:Rusche, Daniel |d 60000 |e 60000PR1297866509 |k 0/60000/ |p 2 | ||
| 998 | |g 1020647353 |a Hesser, Jürgen |m 1020647353:Hesser, Jürgen |d 60000 |d 65200 |e 60000PH1020647353 |e 65200PH1020647353 |k 0/60000/ |k 1/60000/65200/ |p 5 | ||
| 998 | |g 1237754003 |a Runz, Marlen |m 1237754003:Runz, Marlen |d 60000 |d 65200 |e 60000PR1237754003 |e 65200PR1237754003 |k 0/60000/ |k 1/60000/65200/ |p 1 |x j | ||
| 999 | |a KXP-PPN1848823711 |e 4331499693 | ||
| BIB | |a Y | ||
| SER | |a journal | ||
| JSO | |a {"recId":"1848823711","name":{"displayForm":["Marlen Runz, Daniel Rusche, Stefan Schmidt, Martin R. Weihrauch, Jürgen Hesser and Cleo-Aron Weis"]},"physDesc":[{"extent":"10 S."}],"relHost":[{"physDesc":[{"extent":"Online-Ressource"}],"recId":"503328960","pubHistory":["1.2006 -"],"id":{"zdb":["2210518-9"],"eki":["503328960"],"issn":["1746-1596"]},"disp":"Normalization of HE-stained histological images using cycle consistent generative adversarial networksDiagnostic pathology","part":{"year":"2021","pages":"1-10","text":"16(2021) vom: Aug., Artikel-ID 71, Seite 1-10","extent":"10","volume":"16"},"origin":[{"publisher":"BioMed Central","dateIssuedKey":"2006","publisherPlace":"[Erscheinungsort nicht ermittelbar]","dateIssuedDisp":"2006-"}],"note":["Gesehen am 01.08.2019"],"type":{"bibl":"periodical","media":"Online-Ressource"},"title":[{"title":"Diagnostic pathology","title_sort":"Diagnostic pathology"}],"language":["eng"]}],"id":{"eki":["1848823711"],"doi":["10.1186/s13000-021-01126-y"]},"person":[{"family":"Runz","display":"Runz, Marlen","given":"Marlen","role":"aut"},{"family":"Rusche","display":"Rusche, Daniel","role":"aut","given":"Daniel"},{"role":"aut","given":"Stefan","family":"Schmidt","display":"Schmidt, Stefan"},{"family":"Weihrauch","display":"Weihrauch, Martin","given":"Martin","role":"aut"},{"display":"Hesser, Jürgen","family":"Hesser","given":"Jürgen","role":"aut"},{"display":"Weis, Cleo-Aron Thias","family":"Weis","given":"Cleo-Aron Thias","role":"aut"}],"language":["eng"],"origin":[{"dateIssuedKey":"2021","dateIssuedDisp":"06 August 2021"}],"note":["Gesehen am 12.06.2023"],"title":[{"title_sort":"Normalization of HE-stained histological images using cycle consistent generative adversarial networks","title":"Normalization of HE-stained histological images using cycle consistent generative adversarial networks"}],"type":{"bibl":"article-journal","media":"Online-Ressource"}} | ||
| SRT | |a RUNZMARLENNORMALIZAT0620 | ||