A data-driven approach for the partial reconstruction of individual human molar teeth using generative deep learning.

Background and objective: Due to the high prevalence of dental caries, fixed dental restorations are regularly required to restore compromised teeth or replace missing teeth while retaining function and aesthetic appearance. The fabrication of dental restorations, however, remains challenging due to the complexity of the human masticatory system as well as the unique morphology of each individual dentition. Adaptation and reworking are frequently required during the insertion of fixed dental prostheses (FDPs), which increase cost and treatment time. This article proposes a data-driven approach for the partial reconstruction of occlusal surfaces based on a data set that comprises 92 3D mesh files of full dental crown restorations. Methods: A Generative Adversarial Network (GAN) is considered for the given task in view of its ability to represent extensive data sets in an unsupervised manner with a wide variety of applications. Having demonstrated good capabilities in terms of image quality and training stability, StyleGAN-2 has been chosen as the main network for generating the occlusal surfaces. A 2D projection method is proposed in order to generate 2D representations of the provided 3D tooth data set for integration with the StyleGAN architecture. The reconstruction capabilities of the trained network are demonstrated by means of 4 common inlay types using a Bayesian Image Reconstruction method. This involves pre-processing the data in order to extract the necessary information of the tooth preparations required for the used method as well as the modification of the initial reconstruction loss. Results: The reconstruction process yields satisfactory visual and quantitative results for all preparations with a root mean square error (RMSE) ranging from 0.02 mm to 0.18 mm. When compared against a clinical procedure for CAD inlay fabrication, the group of dentists preferred the GAN-based restorations for 3 of the total 4 inlay geometries. Conclusions: This article shows the effectiveness of the StyleGAN architecture with a downstream optimization process for the reconstruction of 4 different inlay geometries. The independence of the reconstruction process and the initial training of the GAN enables the application of the method for arbitrary inlay geometries without time-consuming retraining of the GAN.

Generative deep learning approaches for the design of dental restorations: A narrative review.

OBJECTIVES: This study aims to explore and discuss recent advancements in tooth reconstruction utilizing deep learning (DL) techniques. A review on new DL methodologies in partial and full tooth reconstruction is conducted. DATA/SOURCES: PubMed, Google Scholar, and IEEE Xplore databases were searched for articles from 2003 to 2023. STUDY SELECTION: The review includes 9 articles published from 2018 to 2023. The selected articles showcase novel DL approaches for tooth reconstruction, while those concentrating solely on the application or review of DL methods are excluded. The review shows that data is acquired via intraoral scans or laboratory scans of dental plaster models. Common data representations are depth maps, point clouds, and voxelized point clouds. Reconstructions focus on single teeth, using data from adjacent teeth or the entire jaw. Some articles include antagonist teeth data and features like occlusal grooves and gap distance. Primary network architectures include Generative Adversarial Networks (GANs) and Transformers. Compared to conventional digital methods, DL-based tooth reconstruction reports error rates approximately two times lower. CONCLUSIONS: Generative DL models analyze dental datasets to reconstruct missing teeth by extracting insights into patterns and structures. Through specialized application, these models reconstruct morphologically and functionally sound dental structures, leveraging information from the existing teeth. The reported advancements facilitate the feasibility of DL-based dental crown reconstruction. Beyond GANs and Transformers with point clouds or voxels, recent studies indicate promising outcomes with diffusion-based architectures and innovative data representations like wavelets for 3D shape completion and inference problems. CLINICAL SIGNIFICANCE: Generative network architectures employed in the analysis and reconstruction of dental structures demonstrate notable proficiency. The enhanced accuracy and efficiency of DL-based frameworks hold the potential to enhance clinical outcomes and increase patient satisfaction. The reduced reconstruction times and diminished requirement for manual intervention may lead to cost savings and improved accessibility of dental services.

Preoperative Assessment of Language Dominance through Combined Resting-State and Task-Based Functional Magnetic Resonance Imaging.

Brain lesions in language-related cortical areas remain a challenge in the clinical routine. In recent years, the resting-state fMRI (RS-fMRI) was shown to be a feasible method for preoperative language assessment. The aim of this study was to examine whether language-related resting-state components, which have been obtained using a data-driven independent-component-based identification algorithm, can be supportive in determining language dominance in the left or right hemisphere. Twenty patients suffering from brain lesions close to supposed language-relevant cortical areas were included. RS-fMRI and task-based (TB-fMRI) were performed for the purpose of preoperative language assessment. TB-fMRI included a verb generation task with an appropriate control condition (a syllable switching task) to decompose language-critical and language-supportive processes. Subsequently, the best fitting ICA component for the resting-state language network (RSLN) referential to general linear models (GLMs) of the TB-fMRI (including models with and without linguistic control conditions) was identified using an algorithm based on the Dice index. Thereby, the RSLNs associated with GLMs using a linguistic control condition led to significantly higher laterality indices than GLM baseline contrasts. LIs derived from GLM contrasts with and without control conditions alone did not differ significantly. In general, the results suggest that determining language dominance in the human brain is feasible both with TB-fMRI and RS-fMRI, and in particular, the combination of both approaches yields a higher specificity in preoperative language assessment. Moreover, we can conclude that the choice of the language mapping paradigm is crucial for the mentioned benefits.

Brain Connectivity Studies on Structure-Function Relationships: A Short Survey with an Emphasis on Machine Learning.

This short survey reviews the recent literature on the relationship between the brain structure and its functional dynamics. Imaging techniques such as diffusion tensor imaging (DTI) make it possible to reconstruct axonal fiber tracks and describe the structural connectivity (SC) between brain regions. By measuring fluctuations in neuronal activity, functional magnetic resonance imaging (fMRI) provides insights into the dynamics within this structural network. One key for a better understanding of brain mechanisms is to investigate how these fast dynamics emerge on a relatively stable structural backbone. So far, computational simulations and methods from graph theory have been mainly used for modeling this relationship. Machine learning techniques have already been established in neuroimaging for identifying functionally independent brain networks and classifying pathological brain states. This survey focuses on methods from machine learning, which contribute to our understanding of functional interactions between brain regions and their relation to the underlying anatomical substrate.

A Constrained ICA-EMD Model for Group Level fMRI Analysis.

Independent component analysis (ICA), being a data-driven method, has been shown to be a powerful tool for functional magnetic resonance imaging (fMRI) data analysis. One drawback of this multivariate approach is that it is not, in general, compatible with the analysis of group data. Various techniques have been proposed to overcome this limitation of ICA. In this paper, a novel ICA-based workflow for extracting resting-state networks from fMRI group studies is proposed. An empirical mode decomposition (EMD) is used, in a data-driven manner, to generate reference signals that can be incorporated into a constrained version of ICA (cICA), thereby eliminating the inherent ambiguities of ICA. The results of the proposed workflow are then compared to those obtained by a widely used group ICA approach for fMRI analysis. In this study, we demonstrate that intrinsic modes, extracted by EMD, are suitable to serve as references for cICA. This approach yields typical resting-state patterns that are consistent over subjects. By introducing these reference signals into the ICA, our processing pipeline yields comparable activity patterns across subjects in a mathematically transparent manner. Our approach provides a user-friendly tool to adjust the trade-off between a high similarity across subjects and preserving individual subject features of the independent components.

Frequency-Resolved Dynamic Functional Connectivity Reveals Scale-Stable Features of Connectivity-States.

Investigating temporal variability of functional connectivity is an emerging field in connectomics. Entering dynamic functional connectivity by applying sliding window techniques on resting-state fMRI (rs-fMRI) time courses emerged from this topic. We introduce frequency-resolved dynamic functional connectivity (frdFC) by means of multivariate empirical mode decomposition (MEMD) followed up by filter-bank investigations. In general, we find that MEMD is capable of generating time courses to perform frdFC and we discover that the structure of connectivity-states is robust over frequency scales and even becomes more evident with decreasing frequency. This scale-stability varies with the number of extracted clusters when applying k-means. We find a scale-stability drop-off from k = 4 to k = 5 extracted connectivity-states, which is corroborated by null-models, simulations, theoretical considerations, filter-banks, and scale-adjusted windows. Our filter-bank studies show that filter design is more delicate in the rs-fMRI than in the simulated case. Besides offering a baseline for further frdFC research, we suggest and demonstrate the use of scale-stability as a possible quality criterion for connectivity-state and model selection. We present first evidence showing that connectivity-states are both a multivariate, and a multiscale phenomenon. A data repository of our frequency-resolved time-series is provided.

Spatial Mnemonic Encoding: Theta Power Decreases and Medial Temporal Lobe BOLD Increases Co-Occur during the Usage of the Method of Loci.

The method of loci is one, if not the most, efficient mnemonic encoding strategy. This spatial mnemonic combines the core cognitive processes commonly linked to medial temporal lobe (MTL) activity: spatial and associative memory processes. During such processes, fMRI studies consistently demonstrate MTL activity, while electrophysiological studies have emphasized the important role of theta oscillations (3-8 Hz) in the MTL. However, it is still unknown whether increases or decreases in theta power co-occur with increased BOLD signal in the MTL during memory encoding. To investigate this question, we recorded EEG and fMRI separately, while human participants used the spatial method of loci or the pegword method, a similarly associative but nonspatial mnemonic. The more effective spatial mnemonic induced a pronounced theta power decrease source localized to the left MTL compared with the nonspatial associative mnemonic strategy. This effect was mirrored by BOLD signal increases in the MTL. Successful encoding, irrespective of the strategy used, elicited decreases in left temporal theta power and increases in MTL BOLD activity. This pattern of results suggests a negative relationship between theta power and BOLD signal changes in the MTL during memory encoding and spatial processing. The findings extend the well known negative relation of alpha/beta oscillations and BOLD signals in the cortex to theta oscillations in the MTL.

Perceptual learning in patients with macular degeneration.

Patients with age-related macular degeneration (AMD) or hereditary macular dystrophies (JMD) rely on an efficient use of their peripheral visual field. We trained eight AMD and five JMD patients to perform a texture-discrimination task (TDT) at their preferred retinal locus (PRL) used for fixation. Six training sessions of approximately one hour duration were conducted over a period of approximately 3 weeks. Before, during and after training twelve patients and twelve age-matched controls (the data from two controls had to be discarded later) took part in three functional magnetic resonance imaging (fMRI) sessions to assess training-related changes in the BOLD response in early visual cortex. Patients benefited from the training measurements as indexed by significant decrease (p = 0.001) in the stimulus onset asynchrony (SOA) between the presentation of the texture target on background and the visual mask, and in a significant location specific effect of the PRL with respect to hit rate (p = 0.014). The following trends were observed: (i) improvement in Vernier acuity for an eccentric line-bisection task; (ii) positive correlation between the development of BOLD signals in early visual cortex and initial fixation stability (r = 0.531); (iii) positive correlation between the increase in task performance and initial fixation stability (r = 0.730). The first two trends were non-significant, whereas the third trend was significant at p = 0.014, Bonferroni corrected. Consequently, our exploratory study suggests that training on the TDT can enhance eccentric vision in patients with central vision loss. This enhancement is accompanied by a modest alteration in the BOLD response in early visual cortex.

The effect of feedback on performance and brain activation during perceptual learning.

We investigated the role of informative feedback on the neural correlates of perceptual learning in a coherent-motion detection paradigm. Stimulus displays consisted of four patches of moving dots briefly (500 ms) presented simultaneously, one patch in each visual quadrant. The coherence level was varied in the target patch from near threshold to high, while the other three patches contained only noise. The participants judged whether coherent motion was present or absent in the target patch. To guarantee central fixation, a secondary RSVP digit-detection task was performed at fixation. Over six training sessions subjects learned to detect coherent motion in a predefined quadrant (i.e., the learned location). Half of our subjects were randomly assigned to the feedback group, where they received informative feedback after each response during training, whereas the other group received non-informative feedback during training that a response button was pressed. We investigated whether the presence of informative feedback during training had an influence on the learning success and on the resulting BOLD response. Behavioral data of 24 subjects showed improved performance with increasing practice. Informative feedback promoted learning for motion displays with high coherence levels, whereas it had little effect on learning for displays with near-threshold coherence levels. Learning enhanced fMRI responses in early visual cortex and motion-sensitive area MT+ and these changes were most pronounced for high coherence levels. Activation in the insular and cingulate cortex was mainly influenced by coherence level and trained location. We conclude that feedback modulates behavioral performance and, to a lesser extent, brain activation in areas responsible for monitoring perceptual learning.