Frontiers in Neuroscience (Dec 2021)
Robust, Primitive, and Unsupervised Quality Estimation for Segmentation Ensembles
- Florian Kofler,
- Florian Kofler,
- Florian Kofler,
- Ivan Ezhov,
- Ivan Ezhov,
- Lucas Fidon,
- Carolin M. Pirkl,
- Johannes C. Paetzold,
- Johannes C. Paetzold,
- Egon Burian,
- Sarthak Pati,
- Sarthak Pati,
- Sarthak Pati,
- Sarthak Pati,
- Malek El Husseini,
- Malek El Husseini,
- Fernando Navarro,
- Fernando Navarro,
- Fernando Navarro,
- Suprosanna Shit,
- Suprosanna Shit,
- Jan Kirschke,
- Spyridon Bakas,
- Spyridon Bakas,
- Spyridon Bakas,
- Claus Zimmer,
- Benedikt Wiestler,
- Bjoern H. Menze,
- Bjoern H. Menze
Affiliations
- Florian Kofler
- Department of Informatics, Technical University Munich, Munich, Germany
- Florian Kofler
- Department of Diagnostic and Interventional Neuroradiology, School of Medicine, Klinikum rechts der Isar, Technical University of Munich, Munich, Germany
- Florian Kofler
- TranslaTUM - Central Institute for Translational Cancer Research, Technical University of Munich, Munich, Germany
- Ivan Ezhov
- Department of Informatics, Technical University Munich, Munich, Germany
- Ivan Ezhov
- TranslaTUM - Central Institute for Translational Cancer Research, Technical University of Munich, Munich, Germany
- Lucas Fidon
- School of Biomedical Engineering & Imaging Sciences, King's College London, London, United Kingdom
- Carolin M. Pirkl
- Department of Informatics, Technical University Munich, Munich, Germany
- Johannes C. Paetzold
- Department of Informatics, Technical University Munich, Munich, Germany
- Johannes C. Paetzold
- TranslaTUM - Central Institute for Translational Cancer Research, Technical University of Munich, Munich, Germany
- Egon Burian
- Department of Diagnostic and Interventional Neuroradiology, School of Medicine, Klinikum rechts der Isar, Technical University of Munich, Munich, Germany
- Sarthak Pati
- Department of Informatics, Technical University Munich, Munich, Germany
- Sarthak Pati
- Center for Biomedical Image Computing and Analytics, University of Pennsylvania, Pennsylvania, PA, United States
- Sarthak Pati
- Department of Radiology, Perelman School of Medicine, University of Pennsylvania, Pennsylvania, PA, United States
- Sarthak Pati
- Department of Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania, Pennsylvania, PA, United States
- Malek El Husseini
- Department of Informatics, Technical University Munich, Munich, Germany
- Malek El Husseini
- Department of Diagnostic and Interventional Neuroradiology, School of Medicine, Klinikum rechts der Isar, Technical University of Munich, Munich, Germany
- Fernando Navarro
- Department of Informatics, Technical University Munich, Munich, Germany
- Fernando Navarro
- TranslaTUM - Central Institute for Translational Cancer Research, Technical University of Munich, Munich, Germany
- Fernando Navarro
- Department of Radio Oncology and Radiation Therapy, School of Medicine, Klinikum rechts der Isar, Technical University of Munich, Munich, Germany
- Suprosanna Shit
- Department of Informatics, Technical University Munich, Munich, Germany
- Suprosanna Shit
- TranslaTUM - Central Institute for Translational Cancer Research, Technical University of Munich, Munich, Germany
- Jan Kirschke
- Department of Diagnostic and Interventional Neuroradiology, School of Medicine, Klinikum rechts der Isar, Technical University of Munich, Munich, Germany
- Spyridon Bakas
- Center for Biomedical Image Computing and Analytics, University of Pennsylvania, Pennsylvania, PA, United States
- Spyridon Bakas
- Department of Radiology, Perelman School of Medicine, University of Pennsylvania, Pennsylvania, PA, United States
- Spyridon Bakas
- Department of Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania, Pennsylvania, PA, United States
- Claus Zimmer
- Department of Diagnostic and Interventional Neuroradiology, School of Medicine, Klinikum rechts der Isar, Technical University of Munich, Munich, Germany
- Benedikt Wiestler
- Department of Diagnostic and Interventional Neuroradiology, School of Medicine, Klinikum rechts der Isar, Technical University of Munich, Munich, Germany
- Bjoern H. Menze
- Department of Informatics, Technical University Munich, Munich, Germany
- Bjoern H. Menze
- Department of Quantitative Biomedicine, University of Zurich, Zurich, Switzerland
- DOI
- https://doi.org/10.3389/fnins.2021.752780
- Journal volume & issue
-
Vol. 15
Abstract
A multitude of image-based machine learning segmentation and classification algorithms has recently been proposed, offering diagnostic decision support for the identification and characterization of glioma, Covid-19 and many other diseases. Even though these algorithms often outperform human experts in segmentation tasks, their limited reliability, and in particular the inability to detect failure cases, has hindered translation into clinical practice. To address this major shortcoming, we propose an unsupervised quality estimation method for segmentation ensembles. Our primitive solution examines discord in binary segmentation maps to automatically flag segmentation results that are particularly error-prone and therefore require special assessment by human readers. We validate our method both on segmentation of brain glioma in multi-modal magnetic resonance - and of lung lesions in computer tomography images. Additionally, our method provides an adaptive prioritization mechanism to maximize efficacy in use of human expert time by enabling radiologists to focus on the most difficult, yet important cases while maintaining full diagnostic autonomy. Our method offers an intuitive and reliable uncertainty estimation from segmentation ensembles and thereby closes an important gap toward successful translation of automatic segmentation into clinical routine.
Keywords
