1. Deep learning for the fully automated segmentation of the inner ear on MRI
- Author
-
Tjasse D. Bruintjes, Marly F. J. A. van der Lubbe, Sean Walsh, Benjamin Miraglio, Wim Vos, Sergey Primakov, Alida A. Postma, Raymond van de Berg, Vincent Van Rompaey, Ralph T.H. Leijenaar, Henry C. Woodruff, Philippe Lambin, Akshayaa Vaidyanathan, Patrick F. M. Dammeijer, Marc van Hoof, Monique A. L. Bilderbeek, Hammer Sebastiaan, Fadila Zerka, RS: GROW - R3 - Innovative Cancer Diagnostics & Therapy, Precision Medicine, KNO, RS: MHeNs - R1 - Cognitive Neuropsychiatry and Clinical Neuroscience, MUMC+: MA AIOS Keel Neus Oorheelkunde (9), Beeldvorming, MUMC+: DA BV AIOS Nucleaire Geneeskunde (9), MUMC+: DA BV AIOS Radiologie (9), MUMC+: DA BV Medisch Specialisten Radiologie (9), MUMC+: MA Vestibulogie (9), and MUMC+: MA Keel Neus Oorheelkunde (9)
- Subjects
Adult ,Male ,Similarity (geometry) ,INFORMATION ,Mathematics and computing ,Computer science ,Science ,Datasets as Topic ,Surgical planning ,Article ,030218 nuclear medicine & medical imaging ,03 medical and health sciences ,Deep Learning ,Imaging, Three-Dimensional ,0302 clinical medicine ,Image processing ,medicine ,Humans ,Segmentation ,Inner ear ,Aged ,Multidisciplinary ,business.industry ,Deep learning ,Pattern recognition ,Middle Aged ,Magnetic Resonance Imaging ,Visualization ,medicine.anatomical_structure ,Fully automated ,Ear, Inner ,Feasibility Studies ,Medicine ,Female ,Human medicine ,Artificial intelligence ,RADIOMICS ,business ,Engineering sciences. Technology ,030217 neurology & neurosurgery - Abstract
Segmentation of anatomical structures is valuable in a variety of tasks, including 3D visualization, surgical planning, and quantitative image analysis. Manual segmentation is time-consuming and deals with intra and inter-observer variability. To develop a deep-learning approach for the fully automated segmentation of the inner ear in MRI, a 3D U-net was trained on 944 MRI scans with manually segmented inner ears as reference standard. The model was validated on an independent, multicentric dataset consisting of 177 MRI scans from three different centers. The model was also evaluated on a clinical validation set containing eight MRI scans with severe changes in the morphology of the labyrinth. The 3D U-net model showed precise Dice Similarity Coefficient scores (mean DSC-0.8790) with a high True Positive Rate (91.5%) and low False Discovery Rate and False Negative Rates (14.8% and 8.49% respectively) across images from three different centers. The model proved to perform well with a DSC of 0.8768 on the clinical validation dataset. The proposed auto-segmentation model is equivalent to human readers and is a reliable, consistent, and efficient method for inner ear segmentation, which can be used in a variety of clinical applications such as surgical planning and quantitative image analysis.
- Published
- 2021
- Full Text
- View/download PDF