Retinal OCT Image Segmentation: How Well do Algorithms Generalize or How Transferable are the Data?

Melinščak, Martina; Lončarić, Sven

doi:10.23919/mipro48935.2020.9245336

Cited by 3 publications

(2 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, associating labels to a massive number of images to effectively train a CNN may be extremely problematic in a number of real-world applications. Significant examples are the medical and computational biology domains, where image annotation is an especially cumbersome and time-consuming task that requires solid domain expertise and, more often than not, necessitates consensus strategies to aggregate annotations from several experts to solve class variability problems [2][3][4] . Moreover, biological systems are affected by multiple sources of variability that make the definition of a supervised task impractical, as they require to discover new effects that were not observed during the generation of the training set.…”

Section: Introductionmentioning

confidence: 99%

Exploiting Generative Self-Supervised Learning For The Assessment of Biological Images With Lack of Annotations: A COVID-19 Case-Study

Mascolini

Cardamone

Ponzio

et al. 2021

Preprint

View full text Add to dashboard Cite

Computer-aided analysis of biological images typically requires extensive training on large-scale annotated datasets, which is not viable in many situations. In this paper, we present GAN-DL, a Discriminator Learner based on the StyleGAN2 architecture, which we employ for self-supervised image representation learning in the case of fluorescent biological images. We show that Wasserstein Generative Adversarial Networks combined with linear Support Vector Machines enable high-throughput compound screening based on raw images. We demonstrate this by classifying active and inactive compounds tested for the inhibition of SARS-CoV-2 infection in VERO and HRCE cell lines. In contrast to previous methods, our deep learning-based approach does not require any annotation besides the one that is normally collected during the sample preparation process. We test our technique on the RxRx19a Sars-CoV-2 image collection. The dataset consists of fluorescent images that were generated to assess the ability of regulatory-approved or late-stage clinical trials compounds to modulate the in vitro infection from SARS-CoV-2 in both VERO and HRCE cell lines. We show that our technique can be exploited not only for classification tasks but also to effectively derive a dose-response curve for the tested treatments, in a self-supervised manner. Lastly, we demonstrate its generalization capabilities by successfully addressing a zero-shot learning task, consisting of the categorization of four different cell types of the RxRx1 fluorescent images collection.

show abstract

Section: Introductionmentioning

confidence: 99%

Exploiting Generative Self-Supervised Learning For The Assessment of Biological Images With Lack of Annotations: A COVID-19 Case-Study

Mascolini

Cardamone

Ponzio

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Significant examples are the medical and computational biology domains, where image annotation is an especially cumbersome and time-consuming task that requires solid domain expertise and, more often than not, necessitates consensus strategies to aggregate annotations from several experts to solve class variability problems. [2][3][4] Moreover, biological systems are affected by multiple sources of variability that make the definition of a supervised task impractical, as they require to discover new effects that were not observed during the generation of the training set. On the other hand, a considerable amount of literature focused on machine learning systems, especially CNNs, able to adapt to new conditions without needing a large amount of high-cost data annotations.…”

Section: Introductionmentioning

confidence: 99%

Exploiting generative self-supervised learning for the assessment of biological images with lack of annotations: a COVID-19 case-study

Mascolini¹,

Cardamone²,

Ponzio³

et al. 2021

Preprint

View full text Add to dashboard Cite

Computer-aided analysis of biological images typically requires extensive training on large-scale annotated datasets, which is not viable in many situations. In this paper we present GAN-DL, a Discriminator Learner based on the StyleGAN2 architecture, which we employ for self-supervised image representation learning in the case of fluorescent biological images. We show that Wasserstein Generative Adversarial Networks combined with linear Support Vector Machines enable highthroughput compound screening based on raw images. We demonstrate this by classifying active and inactive compounds tested for the inhibition of SARS-CoV-2 infection in VERO and HRCE cell lines. In contrast to previous methods, our deep learning based approach does not require any annotation besides the one that is normally collected during the sample preparation process. We test our technique on the RxRx19a Sars-CoV-2 image collection. The dataset consists of fluorescent images that were generated to assess the ability of regulatory-approved or in late-stage clinical trials compound to modulate the in vitro infection from SARS-CoV-2 in both VERO and HRCE cell lines. We show that our technique can be exploited not only for classification tasks, but also to effectively derive a dose response curve for the tested treatments, in a self-supervised manner. Lastly, we demonstrate its generalization capabilities by successfully addressing a zero-shot learning task, consisting in the categorization of four different cell types of the RxRx1 fluorescent images collection.

show abstract

Exploiting generative self-supervised learning for the assessment of biological images with lack of annotations

et al. 2022

View full text Add to dashboard Cite

Motivation Computer-aided analysis of biological images typically requires extensive training on large-scale annotated datasets, which is not viable in many situations. In this paper, we present Generative Adversarial Network Discriminator Learner (GAN-DL), a novel self-supervised learning paradigm based on the StyleGAN2 architecture, which we employ for self-supervised image representation learning in the case of fluorescent biological images. Results We show that Wasserstein Generative Adversarial Networks enable high-throughput compound screening based on raw images. We demonstrate this by classifying active and inactive compounds tested for the inhibition of SARS-CoV-2 infection in two different cell models: the primary human renal cortical epithelial cells (HRCE) and the African green monkey kidney epithelial cells (VERO). In contrast to previous methods, our deep learning-based approach does not require any annotation, and can also be used to solve subtle tasks it was not specifically trained on, in a self-supervised manner. For example, it can effectively derive a dose-response curve for the tested treatments. Availability and implementation Our code and embeddings are available at https://gitlab.com/AlesioRFM/gan-dl StyleGAN2 is available at https://github.com/NVlabs/stylegan2.

show abstract

Retinal OCT Image Segmentation: How Well do Algorithms Generalize or How Transferable are the Data?

Cited by 3 publications

References 6 publications

Exploiting Generative Self-Supervised Learning For The Assessment of Biological Images With Lack of Annotations: A COVID-19 Case-Study

Exploiting Generative Self-Supervised Learning For The Assessment of Biological Images With Lack of Annotations: A COVID-19 Case-Study

Exploiting generative self-supervised learning for the assessment of biological images with lack of annotations: a COVID-19 case-study

Exploiting generative self-supervised learning for the assessment of biological images with lack of annotations

Contact Info

Product

Resources

About