f‐AnoGAN: Fast unsupervised anomaly detection with generative adversarial networks

By Thomas Schlegl, Philipp Seeböck, Sebastian M. Waldstein, Georg Langs, Ursula Schmidt-Erfurth. Medical Image Anal., 2019

1,462 citations recorded by the source corpus.

&NA; Obtaining expert labels in clinical imaging is difficult since exhaustive annotation is time‐consuming. Furthermore, not all possibly relevant markers may be known and sufficiently well described a priori to even guide annotation. While supervised learning yields good results if expert labeled training data is available, the visual variability, and thus the vocabulary of findings, we can detect and exploit, is limited to the annotated lesions. Here, we present fast AnoGAN (f‐AnoGAN), a generative adversarial network (GAN) based unsupervised learning approach capable of identifying anomalous images and image segments, that can serve as imaging biomarker candidates. We build a generative model of healthy training data, and propose and evaluate a fast mapping technique of new data to the GAN's latent space. The mapping is based on a trained encoder, and anomalies are detected via a combined anomaly score based on the building blocks of the trained model – comprising a discriminator feature residual error and an image reconstruction error. In the experiments on optical coherence tomography data, we compare the proposed method with alternative approaches, and provide comprehensive empirical evidence that f‐AnoGAN outperforms alternative approaches and yields high anomaly detection accuracy. In addition, a visual Turing test with two retina experts showed that the generated images are indistinguishable from real normal retinal OCT images. The f‐AnoGAN code is available at https://github.com/tSchlegl/f‐AnoGAN.

Subfields: Artificial Intelligence, Biophysics, Radiology, Nuclear Medicine and Imaging

View Paper: https://www.semanticscholar.org/paper/f88cfc38dec02dcf050eb1f56d2d59d90b24e04c

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Biophysics

In Radiology, Nuclear Medicine and Imaging

Citations in the corpus

Surface Defect Detection Methods for Industrial Products: A Review (2021, background): “In 2019, literature [93] improved AnoGAN and proposed the f-AnoGAN model.”
Deep Learning for Medical Anomaly Detection – A Survey (2020, methodology): “al [38], where the authors used a GAN framework for anomaly detection in Optical Coherence Tomography (OCT).”
DRÆM – A discriminatively trained reconstruction embedding for surface anomaly detection (2021, methodology): “In [23] an encoder network is additionally trained that finds the latent representation of the input image that minimizes the reconstruction loss when used as the input by the pretrained generator.”
Mixed supervision for surface-defect detection: from weakly to fully supervised learning (2021, positive): “f-AnoGAN We used publicly the available TensorFlow implementation of f-AnoGAN [32].”
Student-Teacher Feature Pyramid Matching for Anomaly Detection (2021, methodology): “Columns from left to right correspond to input images with defects (ground truth regions in red), anomaly maps of the three blocks, and the resulting anomaly maps respectively. diagnosis [34, 35, 40].”
DeSTSeg: Segmentation Guided Denoising Student-Teacher for Anomaly Detection (2022, methodology): “Visual anomaly detection (AD) with localization is an essential task in many computer vision applications such as industrial inspection [24, 36], medical disease screening [27, 32], and video surveillance [18, 20].”
Multiresolution Knowledge Distillation for Anomaly Detection (2020, methodology): “Due to the unsupervised nature of the AD problem and the restricted data access, availability of just the normal data in training, the majority of methods [36, 31, 40, 18, 34] model the normal data abstraction by extracting semantically meaningful latent features.”
Unsupervised Brain Abnormality Detection Using High Fidelity Image Reconstruction Networks (2020, methodology): “In the context of medical imaging, Schlegl et al. (Schlegl et al., 2017) ﬁrst proposed the AnoGAN framework, which utilizes a deep convolutional GAN (Radford et al., 2016).”
The MVTec Anomaly Detection Dataset: A Comprehensive Real-World Dataset for Unsupervised Anomaly Detection (2021, methodology): “For faster inference, Schlegl et al. (2019) propose to train an additional encoder network that maps input images to their respective latent samples with a single forward pass.”
CFLOW-AD: Real-Time Unsupervised Anomaly Detection with Localization via Conditional Normalizing Flows (2021, methodology): “A fully-generative models [35, 36, 3, 38] are directly applied to images in order to estimate pixel-level probability density and compute per-pixel reconstruction errors as”
Anomaly Detection via Reverse Distillation from One-Class Embedding (2022, methodology): “These tasks include, but not limited to, sample reconstruction [2,5,11,16,26,34,38,48], pseudo-outlier augmenFigure 1.”
Anomaly Transformer: Time Series Anomaly Detection with Association Discrepancy (2021, methodology): “GANs (Goodfellow et al., 2014) are also used for reconstruction-based anomaly detection (Schlegl et al., 2019; Li et al., 2019a; Zhou et al., 2019) and perform as an adversarial regularization.”
Potential, challenges and future directions for deep learning in prognostics and health management applications (2020, neutral): “Recent works by Schlegl et al. (2019) and Ducoffe et al. (2019) combine W-GAN and an encoder to perform anomaly detection on medical images and time series respectively.”
Same Same But DifferNet: Semi-Supervised Defect Detection with Normalizing Flows (2020, methodology): “[26] propose a twostage training method: The GAN is learned first and an encoder is optimized as an inverse generator.”
EfficientAD: Accurate Visual Anomaly Detection at Millisecond-Level Latencies (2023, methodology): “Generative models such as autoencoders [6,8,12,19,31, 41, 52] and GANs [2, 20, 42, 55, 56] have been used extensively for anomaly detection.”
A review of deep learning in medical imaging: Imaging traits, technology trends, case studies with progress highlights, and future promises (2020, background): “[69] build a deep model from only normal images to detect abnormal regions in a test image.”
Deep Learning for Anomaly Detection (2020, methodology): “They may be used to further enhance the anomaly detection performance of the above methods, such as replacing the standard GAN with Wasserstein GAN [137].”