Multiscale generative model using regularized skip-connections and perceptual loss for anomaly detection in toxicologic histopathology
Background: Automated anomaly detection is an important tool that has been developed for many real-world applications, including security systems, industrial inspection, and medical diagnostics. Despite extensive use of machine learning for anomaly detection in these varied contexts, it is challengi...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2022-01-01
|
| Series: | Journal of Pathology Informatics |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2153353922006964 |
| _version_ | 1797977018748370944 |
|---|---|
| author | Philip Zehnder Jeffrey Feng Reina N. Fuji Ruth Sullivan Fangyao Hu |
| author_facet | Philip Zehnder Jeffrey Feng Reina N. Fuji Ruth Sullivan Fangyao Hu |
| author_sort | Philip Zehnder |
| collection | DOAJ |
| description | Background: Automated anomaly detection is an important tool that has been developed for many real-world applications, including security systems, industrial inspection, and medical diagnostics. Despite extensive use of machine learning for anomaly detection in these varied contexts, it is challenging to generalize and apply these methods to complex tasks such as toxicologic histopathology (TOXPATH) assessment (i.e.,finding abnormalities in organ tissues). In this work, we introduce an anomaly detection method using deep learning that greatly improves model generalizability to TOXPATH data. Methods: We evaluated a one-class classification approach that leverages novel regularization and perceptual techniques within generative adversarial network (GAN) and autoencoder architectures to accurately detect anomalous histopathological findings of varying degrees of complexity. We also utilized multiscale contextual data and conducted a thorough ablation study to demonstrate the efficacy of our method. We trained our models on data from normal whole slide images (WSIs) of rat liver sections and validated on WSIs from three anomalous classes. Anomaly scores are collated into heatmaps to localize anomalies within WSIs and provide human-interpretable results. Results: Our method achieves 0.953 area under the receiver operating characteristic on a real-worldTOXPATH dataset. The model also shows good performance at detecting a wide variety of anomalies demonstrating our method’s ability to generalize to TOXPATH data. Conclusion: Anomalies in both TOXPATH histological and non-histological datasets were accurately identified with our method, which was only trained with normal data. |
| first_indexed | 2024-04-11T05:00:18Z |
| format | Article |
| id | doaj.art-defac7b907eb4fb2a6020249fff1de15 |
| institution | Directory Open Access Journal |
| issn | 2153-3539 |
| language | English |
| last_indexed | 2024-04-11T05:00:18Z |
| publishDate | 2022-01-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Journal of Pathology Informatics |
| spelling | doaj.art-defac7b907eb4fb2a6020249fff1de152022-12-26T04:08:40ZengElsevierJournal of Pathology Informatics2153-35392022-01-0113100102Multiscale generative model using regularized skip-connections and perceptual loss for anomaly detection in toxicologic histopathologyPhilip Zehnder0Jeffrey Feng1Reina N. Fuji2Ruth Sullivan3Fangyao Hu4Department of Safety Assessment, Genentech Inc., 1 DNA Way, South San Francisco, CA 94080, USADepartment of Safety Assessment, Genentech Inc., 1 DNA Way, South San Francisco, CA 94080, USADepartment of Safety Assessment, Genentech Inc., 1 DNA Way, South San Francisco, CA 94080, USADepartment of Safety Assessment, Genentech Inc., 1 DNA Way, South San Francisco, CA 94080, USACorresponding author.; Department of Safety Assessment, Genentech Inc., 1 DNA Way, South San Francisco, CA 94080, USABackground: Automated anomaly detection is an important tool that has been developed for many real-world applications, including security systems, industrial inspection, and medical diagnostics. Despite extensive use of machine learning for anomaly detection in these varied contexts, it is challenging to generalize and apply these methods to complex tasks such as toxicologic histopathology (TOXPATH) assessment (i.e.,finding abnormalities in organ tissues). In this work, we introduce an anomaly detection method using deep learning that greatly improves model generalizability to TOXPATH data. Methods: We evaluated a one-class classification approach that leverages novel regularization and perceptual techniques within generative adversarial network (GAN) and autoencoder architectures to accurately detect anomalous histopathological findings of varying degrees of complexity. We also utilized multiscale contextual data and conducted a thorough ablation study to demonstrate the efficacy of our method. We trained our models on data from normal whole slide images (WSIs) of rat liver sections and validated on WSIs from three anomalous classes. Anomaly scores are collated into heatmaps to localize anomalies within WSIs and provide human-interpretable results. Results: Our method achieves 0.953 area under the receiver operating characteristic on a real-worldTOXPATH dataset. The model also shows good performance at detecting a wide variety of anomalies demonstrating our method’s ability to generalize to TOXPATH data. Conclusion: Anomalies in both TOXPATH histological and non-histological datasets were accurately identified with our method, which was only trained with normal data.http://www.sciencedirect.com/science/article/pii/S2153353922006964Anomaly detectionDigital pathologyDeep learningToxicological pathology |
| spellingShingle | Philip Zehnder Jeffrey Feng Reina N. Fuji Ruth Sullivan Fangyao Hu Multiscale generative model using regularized skip-connections and perceptual loss for anomaly detection in toxicologic histopathology Journal of Pathology Informatics Anomaly detection Digital pathology Deep learning Toxicological pathology |
| title | Multiscale generative model using regularized skip-connections and perceptual loss for anomaly detection in toxicologic histopathology |
| title_full | Multiscale generative model using regularized skip-connections and perceptual loss for anomaly detection in toxicologic histopathology |
| title_fullStr | Multiscale generative model using regularized skip-connections and perceptual loss for anomaly detection in toxicologic histopathology |
| title_full_unstemmed | Multiscale generative model using regularized skip-connections and perceptual loss for anomaly detection in toxicologic histopathology |
| title_short | Multiscale generative model using regularized skip-connections and perceptual loss for anomaly detection in toxicologic histopathology |
| title_sort | multiscale generative model using regularized skip connections and perceptual loss for anomaly detection in toxicologic histopathology |
| topic | Anomaly detection Digital pathology Deep learning Toxicological pathology |
| url | http://www.sciencedirect.com/science/article/pii/S2153353922006964 |
| work_keys_str_mv | AT philipzehnder multiscalegenerativemodelusingregularizedskipconnectionsandperceptuallossforanomalydetectionintoxicologichistopathology AT jeffreyfeng multiscalegenerativemodelusingregularizedskipconnectionsandperceptuallossforanomalydetectionintoxicologichistopathology AT reinanfuji multiscalegenerativemodelusingregularizedskipconnectionsandperceptuallossforanomalydetectionintoxicologichistopathology AT ruthsullivan multiscalegenerativemodelusingregularizedskipconnectionsandperceptuallossforanomalydetectionintoxicologichistopathology AT fangyaohu multiscalegenerativemodelusingregularizedskipconnectionsandperceptuallossforanomalydetectionintoxicologichistopathology |