ABSTRACT
Automated analysis of the inner ear anatomy in radiological data instead of time-consuming manual assessment is a worthwhile goal that could facilitate preoperative planning and clinical research. We propose a framework encompassing joint semantic segmentation of the inner ear and anatomical landmark detection of helicotrema, oval and round window. A fully automated pipeline with a single, dual-headed volumetric 3D U-Net was implemented, trained and evaluated using manually labeled in-house datasets from cadaveric specimen ([Formula: see text]) and clinical practice ([Formula: see text]). The model robustness was further evaluated on three independent open-source datasets ([Formula: see text] scans) consisting of cadaveric specimen scans. For the in-house datasets, Dice scores of [Formula: see text], intersection-over-union scores of [Formula: see text] and average Hausdorff distances of [Formula: see text] and [Formula: see text] voxel units were achieved. The landmark localization task was performed automatically with an average localization error of [Formula: see text] voxel units. A robust, albeit reduced performance could be attained for the catalogue of three open-source datasets. Results of the ablation studies with 43 mono-parametric variations of the basal architecture and training protocol provided task-optimal parameters for both categories. Ablation studies against single-task variants of the basal architecture showed a clear performance benefit of coupling landmark localization with segmentation and a dataset-dependent performance impact on segmentation ability.
Subject(s)
Deep Learning , Ear, Inner , Humans , Ear, Inner/diagnostic imaging , Cadaver , Image Processing, Computer-Assisted/methodsABSTRACT
In order to treat degenerative diseases, the importance of advanced therapy medicinal products has increased in recent years. The newly developed treatment strategies require a rethinking of the appropriate analytical methods. Current standards are missing the complete and sterile analysis of the product of interest to make the drug manufacturing effort worthwhile. They only consider partial areas of the sample or product while also irreversibly damaging the investigated specimen. Two-dimensional T1 / T2 MR relaxometry meets these requirements and is therefore a promising in-process control during the manufacturing and classification process of cell-based treatments. In this study a tabletop MR scanner was used to perform two-dimensional MR relaxometry. Throughput was increased by developing an automation platform based on a low-cost robotic arm, resulting in the acquisition of a large dataset of cell-based measurements. Two-dimensional inverse Laplace transformation was used for post-processing, followed by data classification performed with support vector machines (SVM) as well as optimized artificial neural networks (ANN). The trained networks were able to distinguish non-differentiated from differentiated MSCs with a prediction accuracy of 85%. To increase versatility, an ANN was trained on 354 independent, biological replicates distributed across ten different cell lines, resulting in a prediction accuracy of up to 98% depending on data composition. The present study provides a proof of principle for the application of T1 / T2 relaxometry as a non-destructive cell classification method. It does not require labeling of cells and can perform whole mount analysis of each sample. Since all measurements can be performed under sterile conditions, it can be used as an in-process control for cellular differentiation. This distinguishes it from other characterization techniques, as most are destructive or require some type of cell labeling. These advantages highlight the technique's potential for preclinical screening of patient-specific cell-based transplants and drugs.
Subject(s)
Artificial Intelligence , Magnetic Resonance Imaging , Humans , Magnetic Resonance Spectroscopy , Neural Networks, Computer , AutomationABSTRACT
PURPOSE: To evaluate an iterative learning approach for enhanced performance of robust artificial-neural-networks for k-space interpolation (RAKI), when only a limited amount of training data (auto-calibration signals [ACS]) are available for accelerated standard 2D imaging. METHODS: In a first step, the RAKI model was tailored for the case of limited training data amount. In the iterative learning approach (termed iterative RAKI [iRAKI]), the tailored RAKI model is initially trained using original and augmented ACS obtained from a linear parallel imaging reconstruction. Subsequently, the RAKI convolution filters are refined iteratively using original and augmented ACS extracted from the previous RAKI reconstruction. Evaluation was carried out on 200 retrospectively undersampled in vivo datasets from the fastMRI neuro database with different contrast settings. RESULTS: For limited training data (18 and 22 ACS lines for R = 4 and R = 5, respectively), iRAKI outperforms standard RAKI by reducing residual artifacts and yields better noise suppression when compared to standard parallel imaging, underlined by quantitative reconstruction quality metrics. Additionally, iRAKI shows better performance than both GRAPPA and standard RAKI in case of pre-scan calibration with varying contrast between training- and undersampled data. CONCLUSION: RAKI benefits from the iterative learning approach, which preserves the noise suppression feature, but requires less original training data for the accurate reconstruction of standard 2D images thereby improving net acceleration.