Exploring the methodological approaches of studies on radiographic databases used in cariology to feed AI: A Systematic Review.

INTRODUCTION: A growing number of studies on diagnostic imaging show superior efficiency and accuracy of computer-aided diagnostic systems compared to certified dentists. This methodological systematic review aims to evaluate the different methodological approaches used by studies focusing on machine learning and deep learning and that have used radiographic databases to classify, detect, and segment dental caries. METHODS: The protocol was registered in PROSPERO before data collection (CRD42022348097). Literature research was performed in MEDLINE, Embase, IEEE Xplore, and Web of Science until December 2022, without language restrictions. Studies and surveys using a dental radiographic database for the classification, detection, or segmentation of carious lesions were sought. Records deemed eligible were retrieved and further assessed for inclusion by two reviewers who resolved any discrepancies through consensus. A third reviewer was consulted when any disagreements or discrepancies persist between the two reviewers. After data extraction, the same reviewers assessed the methodological quality using the CLAIM and QUADAS-AI checklists. RESULTS: After screening 325 articles, 35 studies were eligible and included. The bitewing was the most commonly used radiograph (n=17) at the time when detection (n=15) was the most explored computer vision task. The sample sizes used ranged from 95 to 38437, while the augmented training set ranged from 300 to 315786. Convolutional neural network (CNN) was the most commonly used model. The mean completeness of CLAIM items was 49 % (SD ± 34%). The applicability of the CLAIM checklist items revealed several weaknesses in the methodology of the selected studies: most of the studies were monocentric, and only 9% of them used an external test set when evaluating the model's performance. The QUADAS-AI tool revealed that only 43% of the studies included in this systematic review were at low risk of bias concerning the standard reference domain. CONCLUSION: This review demonstrates that the overall scientific quality of studies conducted to feed AI algorithms is low. Some improvement in the design and validation of studies can be made with the development of a standardized guideline for the reproducibility and generalizability of results and, thus, their clinical applications.

Ultrafast Cardiac Imaging Using Deep Learning for Speckle-Tracking Echocardiography.

High-quality ultrafast ultrasound imaging is based on coherent compounding from multiple transmissions of plane waves (PW) or diverging waves (DW). However, compounding results in reduced frame rate, as well as destructive interferences from high-velocity tissue motion if motion compensation (MoCo) is not considered. While many studies have recently shown the interest of deep learning for the reconstruction of high-quality static images from PW or DW, its ability to achieve such performance while maintaining the capability of tracking cardiac motion has yet to be assessed. In this article, we addressed such issue by deploying a complex-weighted convolutional neural network (CNN) for image reconstruction and a state-of-the-art speckle-tracking method. The evaluation of this approach was first performed by designing an adapted simulation framework, which provides specific reference data, i.e., high-quality, motion artifact-free cardiac images. The obtained results showed that, while using only three DWs as input, the CNN-based approach yielded an image quality and a motion accuracy equivalent to those obtained by compounding 31 DWs free of motion artifacts. The performance was then further evaluated on nonsimulated, experimental in vitro data, using a spinning disk phantom. This experiment demonstrated that our approach yielded high-quality image reconstruction and motion estimation, under a large range of velocities and outperforms a state-of-the-art MoCo-based approach at high velocities. Our method was finally assessed on in vivo datasets and showed consistent improvement in image quality and motion estimation compared to standard compounding. This demonstrates the feasibility and effectiveness of deep learning reconstruction for ultrafast speckle-tracking echocardiography.

Short echo time dual-frequency MR Elastography with Optimal Control RF pulses.

Magnetic Resonance Elastography (MRE) quantifies the mechanical properties of tissues, typically applying motion encoding gradients (MEG). Multifrequency results allow better characterizations of tissues using data usually acquired through sequential monofrequency experiments. High frequencies are difficult to reach due to slew rate limitations and low frequencies induce long TEs, yielding magnitude images with low SNR. We propose a novel strategy to perform simultaneous multifrequency MRE in the absence of MEGs: using RF pulses designed via the Optimal Control (OC) theory. Such pulses control the spatial distribution of the MRI magnetization phase so that the resulting transverse magnetization reproduces the phase pattern of an MRE acquisition. The pulse is applied with a constant gradient during the multifrequency mechanical excitation to simultaneously achieve slice selection and motion encoding. The phase offset sampling strategy can be adapted according to the excitation frequencies to reduce the acquisition time. Phantom experiments were run to compare the classical monofrequency MRE to the OC based dual-frequency MRE method and showed excellent agreement between the reconstructed shear storage modulus G'. Our method could be applied to simultaneously acquire low and high frequency components, which are difficult to encode with the classical MEG MRE strategy.

Complex Convolutional Neural Networks for Ultrafast Ultrasound Imaging Reconstruction From In-Phase/Quadrature Signal.

Ultrafast ultrasound imaging remains an active area of interest in the ultrasound community due to its ultrahigh frame rates. Recently, a wide variety of studies based on deep learning have sought to improve ultrafast ultrasound imaging. Most of these approaches have been performed on radio frequency (RF) signals. However, in- phase/quadrature (I/Q) digital beamformers are now widely used as low-cost strategies. In this work, we used complex convolutional neural networks for reconstruction of ultrasound images from I/Q signals. We recently described a convolutional neural network architecture called ID-Net, which exploited an inception layer designed for reconstruction of RF diverging-wave ultrasound images. In the present study, we derive the complex equivalent of this network, i.e., complex-valued inception for diverging-wave network (CID-Net) that operates on I/Q data. We provide experimental evidence that CID-Net provides the same image quality as that obtained from RF-trained convolutional neural networks, i.e., using only three I/Q images, CID-Net produces high-quality images that can compete with those obtained by coherently compounding 31 RF images. Moreover, we show that CID-Net outperforms the straightforward architecture that consists of processing real and imaginary parts of the I/Q signal separately, which thereby indicates the importance of consistently processing the I/Q signals using a network that exploits the complex nature of such signals.

Harmonic wideband simultaneous dual-frequency MR Elastography.

Magnetic resonance elastography (MRE) is used to non-invasively quantify viscoelastic properties of tissues based on the measurement of propagation characteristics of shear waves. Because some of these viscoelastic parameters show a frequency dependence, multifrequency analysis allows us to measure the wave propagation dispersion, leading to a better characterization of tissue properties. Conventionally, motion encoding gradients (MEGs) oscillating at the same frequency as the mechanical excitation encode motion. Hence, multifrequency data is usually obtained by sequentially repeating monochromatic wave excitations experiments at different frequencies. The result is that the total acquisition time is multiplied by a factor corresponding to the number of repetitions of monofrequency experiments, which is a major limitation of multifrequency MRE. In order to make it more accessible, a novel single-shot harmonic wideband dual-frequency MRE method is proposed. Two superposed shear waves of different frequencies are simultaneously generated and propagate in a sample. Trapezoidal oscillating MEGs are used to encode mechanical vibrations having frequencies that are an odd multiple of the MEG frequency. The number of phase offsets is optimized to reduce the acquisition time. For this purpose, a sampling method not respecting the Shannon theorem is used to produce a controlled temporal aliasing that allows us to encode both frequencies without any additional examination time. Phantom experiments were run to compare conventional monofrequency MRE with the single-shot dual-frequency MRE method and showed excellent agreement between the reconstructed shear storage moduli G'. In addition, dual-frequency MRE yielded an increased signal-to-noise ratio compared with conventional monofrequency MRE acquisitions when encoding the high frequency component. The novel proposed multifrequency MRE method could be applied to simultaneously acquire more than two frequency components, reducing examination time. Further studies are needed to confirm its applicability in preclinical and clinical models.

Reconstruction for Diverging-Wave Imaging Using Deep Convolutional Neural Networks.

In recent years, diverging wave (DW) ultrasound imaging has become a very promising methodology for cardiovascular imaging due to its high temporal resolution. However, if they are limited in number, DW transmits provide lower image quality compared with classical focused schemes. A conventional reconstruction approach consists in summing series of ultrasound signals coherently, at the expense of frame rate, data volume, and computation time. To deal with this limitation, we propose a convolutional neural network (CNN) architecture, Inception for DW Network (IDNet), for high-quality reconstruction of DW ultrasound images using a small number of transmissions. In order to cope with the specificities induced by the sectorial geometry associated with DW imaging, we adopted the inception model composed of the concatenation of multiscale convolution kernels. Incorporating inception modules aims at capturing different image features with multiscale receptive fields. A mapping between low-quality images and corresponding high-quality compounded reconstruction was learned by training the network using in vitro and in vivo samples. The performance of the proposed approach was evaluated in terms of contrast ratio (CR), contrast-to-noise ratio (CNR), and lateral resolution (LR), and compared with standard compounding method and conventional CNN methods. The results demonstrated that our method could produce high-quality images using only 3 DWs, yielding an image quality equivalent to that obtained with compounding of 31 DWs and outperforming more conventional CNN architectures in terms of complexity, inference time, and image quality.

High-Quality Plane Wave Compounding Using Convolutional Neural Networks.

Single plane wave (PW) imaging produces ultrasound images of poor quality at high frame rates (ultrafast). High-quality PW imaging usually relies on the coherent compounding of several successive steered emissions (typically more than ten), which in turn results in a decreased frame rate. We propose a new strategy to reduce the number of emitted PWs by learning a compounding operation from data, i.e., by training a convolutional neural network to reconstruct high-quality images using a small number of transmissions. We present experimental evidence that this approach is promising, as we were able to produce high-quality images from only three PWs, competing in terms of contrast ratio and lateral resolution with the standard compounding of 31 PWs ( 10× speedup factor).

Discrimination between emboli and artifacts for outpatient transcranial Doppler ultrasound data.

This paper addresses the detection of emboli in transcranial Doppler ultrasound data acquired from an original portable device. The challenge is the removal of several artifacts (motion and voice) intrinsically related to long-duration (up to 1 h 40 mn per patient) outpatient signals monitoring from this device, as well as high intensities due to the stochastic nature of blood flow. This paper proposes an adapted removal procedure. This firstly consists of reducing the background noise and detecting the blood flow in the time-frequency domain using a likelihood method for contour detection. Then, a hierarchical extraction of features from magnitude and bounding boxes is achieved for the discrimination of emboli and artifacts. After processing of the long-duration outpatient signals, the number of artifacts predicted as emboli is considerably reduced (by 92% for some parameter values) between the first and the last step of our algorithm.