Deep Learning-based Algorithm Research Articles

Appearance-Based Gaze Estimation With Deep Learning: A Review and Benchmark.

Human gaze provides valuable information on human focus and intentions, making it a crucial area of research. Recently, deep learning has revolutionized appearance-based gaze estimation. However, due to the unique features of gaze estimation research, such as the unfair comparison between 2D gaze positions and 3D gaze vectors and the different pre-processing and post-processing methods, there is a lack of a definitive guideline for developing deep learning-based gaze estimation algorithms. In this paper, we present a systematic review of the appearance-based gaze estimation methods using deep learning. First, we survey the existing gaze estimation algorithms along the typical gaze estimation pipeline: deep feature extraction, deep learning model design, personal calibration and platforms. Second, to fairly compare the performance of different approaches, we summarize the data pre-processing and post-processing methods, including face/eye detection, data rectification, 2D/3D gaze conversion and gaze origin conversion. Finally, we set up a comprehensive benchmark for deep learning-based gaze estimation. We characterize all the public datasets and provide the source code of typical gaze estimation algorithms. This paper serves not only as a reference to develop deep learning-based gaze estimation methods, but also a guideline for future gaze estimation research.

Deep learning-based algorithm for automatic quantification of nigrosome-1 and Parkinsonism classification using susceptibility map-weighted MRI.

To develop and validate a deep learning-based automatic quantification for nigral hyperintensity and a classification algorithm for neurodegenerative parkinsonism using susceptibility map-weighted imaging (SMwI). We retrospectively collected 450 participants (210 with idiopathic Parkinson's disease [IPD] and 240 individuals in the control group) for training data between November 2022 and May 2023, and 237 participants (168 with IPD, 58 with essential tremor, and 11 with drug-induced Parkinsonism) for validation data between July 2021 and January 2022. SMwI data were reconstructed from multi-echo GRE. Diagnostic performance for diagnosing IPD was assessed using deep learning-based automatic quantification (Heuron NI) and classification (Heuron IPD) model. Reference standard for IPD was based on 18F-FP-CIT PET finding. Additionally, the correlation between the H&Y stage and volume of nigral hyperintensity in patients with IPD was assessed. Quantification of nigral hyperintensity using Heuron NI showed AUC of 0.915 (95% CI, 0.872-0.947) and 0.928 (95% CI, 0.887-0.957) on the left and right, respectively. Classification of nigral hyperintensity abnormality using Heuron IPD showed AUC of 0.967 (95% CI, 0.881-0.991) and 0.976 (95% CI, 0.948-0.992) on the left and right, respectively. H&Y score ≥ 3 showed significant smaller nigral hyperintensity volume (1.43 ± 1.19 mm3) compared to H&Y score 1-2.5 (1.98 ± 1.63 mm3; p = 0.008). Our deep learning-based model proves rapid, accurate automatic quantification of nigral hyperintensity, facilitating IPD diagnosis, symptom severity prediction, and patient stratification for personalized therapy. Further study is warranted to validate the findings across various clinical settings. IPD = Idiopathic Parkinson's disease; SN = substantia nigra; SMwI = susceptibility map weighted imaging; QSM = quantitative susceptibility mapping; CNN = convolutional neural network; ICV = intracranial volume.

Advancements in Fingerprint Recognition Through Deep Learning: A Comprehensive Analysis of Novel Algorithms

Abstract. This paper presents a comprehensive exploration of the application of deep learning technology in fingerprint recognition, focusing on the development and impact of several novel algorithms. These algorithms have been specifically designed to address challenges in key sub-fields such as pose estimation, direction field estimation, minutiae extraction, and minutiae matching. By integrating deep learning techniques, these new approaches significantly enhance the accuracy, stability, and efficiency of fingerprint recognition systems. The study demonstrates that these algorithms surpass traditional methods in several critical areas, offering improved precision in recognizing fingerprints, particularly in high-noise environments. Furthermore, the fully differentiable nature of these models contributes to their robustness, enabling more consistent and reliable performance across diverse scenarios. The results underscore the potential for these deep learning-based algorithms to set new benchmarks in the field, with broad implications for their application in security, law enforcement, and other areas requiring reliable biometric authentication. The use of these cutting-edge methods is anticipated to be vital in influencing the direction of fingerprint recognition technology as it develops further, guaranteeing increased security and precision.

Repeatability of Microperimetry in areas of RPE and Photoreceptor loss in Geographic Atrophy supported by AI-based OCT biomarker quantification.

Growing interest in microperimetry (MP) or fundus-controlled perimetry (FCP) as targeted psychometric testing method in geographic atrophy (GA) is warranted due to the disease subclinical/extra-foveal appearance or preexisting foveal loss with visual acuity becoming unreliable. We provide comprehensive pointwise test-retest repeatability reference values on the most widely used MP devices and combine them with targeted testing in areas of retinal pigment epithelium (RPE) as well as photoreceptor (PR) integrity loss, guiding the interpretation of sensitivity loss during the long-term follow-up of GA patients. Prospective reliability study METHODS: Patients with GA underwent consecutive testing on CenterVue (iCare) MAIA and NIDEK MP3 devices. Obtained PWS measurements were spatially co-registered to an optical coherence tomography (OCT) volume scan acquired during the same visit. Areas with RPE and PR integrity loss, drusen and PR thickness as well as the volume of hyperreflective foci (HRF) where identified and quantified using a set of validated deep learning-based algorithms. Test-retest repeatability was assessed according to areas defined by biomarker-specific morphologic changes using Bland-Altmann coefficients of repeatability (CoR). Furthermore, the inter-device correlation, the repeatability of scotoma point detection as well as any potential effects on fixation stability were assessed. 900 stimuli per device from twenty subjects were included. Identical overall PWS test-retest variance could be detected for MAIA (±6.57) and MP3 (±6.59). PR integrity loss was associated with a higher test-retest variance on both devices (MAIA: p=0.002; MP3: p<0.001). Higher CoR for stimuli in areas presenting RPE loss (±10.99 vs ±5.34) or HRF (±9.21 vs ±6.25) could only be detected on MP3 examinations (p<0.001 and p=0.01, respectively). An excellent intra-device correlation (MAIA: 0.94[0.93-0.95] MP3: 0.94[0.94-0.95]) and a good mean inter-device correlation (0.84[0.53-0.92]) could be demonstrated. The chosen device, run order or absence of foveal sparing had no significant effect on fixation stability. Areas presenting automatically quantified PR integrity loss with and without underlying RPE loss are associated with higher test-retest variance for both MAIA and MP3. These findings are crucial for an accurate interpretation of GA progression during long-term follow-up and the planning of future trials with microperimetry testing as functional study endpoint.

Cancer-Associated Lymphoid Aggregates in Histology Images: Manual and Deep Learning-Based Quantification Approaches.

Quantification of lymphoid aggregates including tertiary lymphoid structures (TLS) with germinal centers in histology images of cancer is a promising approach for developing prognostic and predictive tissue biomarkers. In this article, we provide recommendations for identifying lymphoid aggregates in tissue sections from routine pathology workflows such as hematoxylin and eosin staining. To overcome the intrinsic variability associated with manual image analysis (such as subjective decision-making, attention span), we recently developed a deep learning-based algorithm called HookNet-TLS to detect lymphoid aggregates and germinal centers in various tissues. Here, we additionally provide a guideline for using manually annotated images for training and implementing HookNet-TLS for automated and objective quantification of lymphoid aggregates in various cancer types.

PATH-50. AI-POWERED AUTOMATED TISSUE SEGMENTATION IMPROVES OUTCOME STRATIFICATION IN GLIOBLASTOMA

Abstract Glioblastoma displays morphological heterogeneity on routine H&amp;E whole slide images, with conflicting evidence on the reliability of morphological features in predicting tumor behavior and treatment response. However, current standards for assessing morphology and disease status rely on nonquantitative manual evaluation by pathologists, which are prone to inherent variation. We sought to (1) expand the tools available for quantitative assessment and (2) evaluate the utility of this approach in identifying valuable biomarkers. We trained a deep learning-based algorithm using DenseNet on HALO AI platform for automated segmentation of six tissue classes (heavy infiltration, mild infiltration, mesenchymal tissue, geographic necrosis, palisading necrosis, and hemorrhage) on 3232 labeled annotations from 100 whole slide images. Manual pathologists’ estimations from clinical records showed strong correlation with corresponding digital quantifications (p&amp;lt;0.001), and further pixel-wise validation on 828 separate annotations showed high overlap between predicted and ground truth segmentations (precision=0.88, recall= 0.85, F1-score=0.87). We then identified a longitudinal cohort of 115 newly diagnosed glioblastoma tissue paired with the recurrence post-treatment (RT/TMZ), including all 827 H&amp;E whole slide images scanned at 40x. The quantification of longitudinally paired tissue showed a relative increase in geographic necrosis (+62%) and mild infiltration (+110%) and a relative decrease in palisading necrosis (-54%) and heavy infiltration (-35%) between the two timepoints. Patients with rapid clinical progression had more areas of large geographic necrosis (p=0.002), and mesenchymal tissue increase was associated with worse overall survival (p=0.016). Digitally quantified heavy infiltration as a predominant tissue on recurrence outperformed corresponding pathologist estimation of tumor as a predominant tissue on recurrence in stratifying overall-survival (p=0.005 vs p=0.640). Collectively, our AI-powered classifier enables an automated objective quantification of glioblastoma tissue, identifies markers of clinical response, and offers a potential adjunct to enhance routine pathologic reporting of glioblastoma.

A Radio Frequency Interference Screening Framework—From Quick-Look Detection Using Statistics-Assisted Network to Raw Echo Tracing

Synthetic aperture radar (SAR) is often affected by other high-power electromagnetic devices during ground observation, which causes unintentional radio frequency interference (RFI) with the acquired echo, bringing adverse effects into data processing and image interpretation. When faced with the task of screening massive SAR data, there is an urgent need for the global perception and detection of interference. The existing RFI detection method usually only uses a single type of data for detection, ignoring the information association between the data at all levels of the real SAR product, resulting in some computational redundancy. Meanwhile, current deep learning-based algorithms are often unable to locate the range of RFI coverage in the azimuth direction. Therefore, a novel RFI processing framework from quick-looks to single-look complex (SLC) data and then to raw echo is proposed. We take the data of Sentinel-1 terrain observation with progressive scan (TOPS) mode as an example. By combining the statistics-assisted network with the sliding-window algorithm and the error-tolerant training strategy, it is possible to accurately detect and locate RFI in the quick looks of an SLC product. Then, through the analysis of the TOPSAR imaging principle, the position of the RFI in the SLC image is preliminarily confirmed. The possible distribution of the RFI in the corresponding raw echo is further inferred, which is one of the first attempts to use spaceborne SAR data to elucidate the RFI location mapping relationship between image data and raw echo. Compared with directly detecting all of the SLC data, the time for the proposed framework to determine the RFI distribution in the SLC data can be shortened by 53.526%. All the research in this paper is conducted on Sentinel-1 real data, which verify the feasibility and effectiveness of the proposed framework for radio frequency signals monitoring in advanced spaceborne SAR systems.

DeepFake Image Detection: Fake Image Detection using CNNs and GANs Algorithm

Deep learning is a powerful and versatile technique that has seen extensive applications in areas such as natural language processing, machine learning, and computer vision. Among its most recent applications is the generation of deepfakes, which are high-quality, realistic altered videos or images that have garnered significant attention. While innovative uses of deepfake technology are being explored, its potential for misuse has raised serious concerns. Harmful applications, such as spreading fake news, creating celebrity pornography, financial fraud, and revenge pornography, have become increasingly prevalent in the digital age. As a result, public figures, including celebrities and politicians, face heightened risks from deepfake content. In response, substantial research has been conducted to explore the mechanics behind deepfakes, leading to the development of various deep learning-based algorithms for their detection. This study provides a comprehensive review of deepfake creation and detection techniques, focusing on different deep learning approaches. Additionally, it discusses the limitations of existing methods and the availability of datasets for research. The lack of a highly accurate and fully automated deepfake detection system presents a significant challenge as the ease of generating and distributing such content continues to grow. Nonetheless, recent efforts in deep learning have shown promising results, surpassing traditional detection methods.

Prediction of freeze–thaw and thermal shock weathering on natural stones through deep learning-based algorithms

Prediction of freeze–thaw and thermal shock weathering on natural stones through deep learning-based algorithms

A Refined and Efficient CNN Algorithm for Remote Sensing Object Detection

Remote sensing object detection (RSOD) plays a crucial role in resource utilization, geological disaster risk assessment and urban planning. Deep learning-based object-detection algorithms have proven effective in remote sensing image studies. However, accurate detection of objects with small size, dense distribution and complex object arrangement remains a significant challenge in the remote sensing field. To address this, a refined and efficient object-detection algorithm (RE-YOLO) has been proposed in this paper for remote sensing images. Initially, a refined and efficient module (REM) was designed to balance computational complexity and feature-extraction capabilities, which serves as a key component of the RE_CSP block. RE_CSP block efficiently extracts multi-scale information, overcoming challenges posed by complex backgrounds. Moreover, the spatial extracted attention module (SEAM) has been proposed in the bottleneck of backbone to promote representative feature learning and enhance the semantic information capture. In addition, a three-branch path aggregation network (TBPAN) has been constructed as the neck network, which facilitates comprehensive fusion of shallow positional information and deep semantic information across different channels, enabling the network with a robust ability to capture contextual information. Extensive experiments conducted on two large-scale remote sensing datasets, DOTA-v1.0 and SCERL, demonstrate that the proposed RE-YOLO outperforms state-of-the-art other object-detection approaches and exhibits a significant improvement in generalization ability.

ACORN+: Adaptive Compression-Reconstruction for Device-Cloud Collaboration Video Services

With the improvement of edge-based autonomous systems such as mobile Industrial IoT(IIoT) networks, edge devices can capture and upload videos with increasing bitrates. Massive edge-computing end nodes are eager for adequate multimedia data to satisfy the requirements of real-time video services. However, existing encoding standards for video services in Web 2.0 are specifically designed for something other than IoT video streaming. We have improved our Adaptive Compression-Reconstruction framework ACORN to obtain ACORN+, based on compressed sensing and recent advances in deep learning. At end nodes, we compress multiple sequential video frames into a single frame to reduce video volume. Given that multiple kinds of intelligent tasks are expected to be finished on the device side, we also designed a device-cloud collaboration scheme where deep learning-based algorithms can be executed on both the device and server sides. Experiments reveal that video analytics can be conducted on compressed frames. Taking action recognition as a device-cloud collaboration use case, we find ACORN+ obtains more than 3x speedup on compressed frames. The reconstruction algorithm in ACORN+ is with 1- 4 dB improvements. Moreover, the encoding time cost and the encoded video volume are reduced by more than 4x under the ACORN+ framework. 1

Harnessing Deep Learning for Enhanced MPPT in Solar PV Systems: An LSTM Approach Using Real-World Data

Maximum Power Point Tracking (MPPT) is essential for maximizing the efficiency of solar photovoltaic (PV) systems. While numerous MPPT methods exist, practical implementations often lean towards conventional techniques due to their simplicity. However, these traditional methods can struggle with rapid fluctuations in solar irradiance and temperature. This paper introduces a novel deep learning-based MPPT algorithm that leverages a Long Short-Term Memory (LSTM) deep neural network (DNN) to effectively track maximum power from solar PV panels, utilizing real-world data. The simulations of three algorithms—Perturb and Observe (P&amp;O), Artificial Neural Network (ANN), and the proposed LSTM-based MPPT—were conducted using MATLAB (2021b) and RT_LAB (24.3.3) with an OPAL-RT simulator for real-time analysis. The data used for this study were sourced from NASA/POWER’s Native Resolution Daily Data of solar irradiation and temperature specific to Imphal, Manipur, India. The obtained results demonstrate that the LSTM-based MPPT system achieves a superior power tracking accuracy under changing solar conditions, producing an average output of 74 W. In comparison, the ANN and P&amp;O methods yield average outputs of 57 W and 62 W, respectively. This significant improvement, i.e., 20–30%, underscores the effectiveness of the LSTM technique in enhancing the power output of solar PV systems. By incorporating real-world data, valuable insights into solar power generation specific to the selected location are provided. Furthermore, the outputs of the model were verified through real-time simulations using the OPAL-RT simulator OP4510, showcasing the practical applicability of this approach in real-world scenarios.

A multimodal deep learning-based algorithm for specific fetal heart rate events detection.

This study aims to develop a multimodal deep learning-based algorithm for detecting specific fetal heart rate (FHR) events, to enhance automatic monitoring and intelligent assessment of fetal well-being. We analyzed FHR and uterine contraction signals by combining various feature extraction techniques, including morphological features, heart rate variability features, and nonlinear domain features, with deep learning algorithms. This approach enabled us to classify four specific FHR events (bradycardia, tachycardia, acceleration, and deceleration) as well as four distinct deceleration patterns (early, late, variable, and prolonged deceleration). We proposed a multi-model deep neural network and a pre-fusion deep learning model to accurately classify the multimodal parameters derived from Cardiotocography signals. These accuracy metrics were calculated based on expert-labeled data. The algorithm achieved a classification accuracy of 96.2 % for acceleration, 94.4 % for deceleration, 90.9 % for tachycardia, and 85.8 % for bradycardia. Additionally, it achieved 67.0 % accuracy in classifying the four distinct deceleration patterns, with 80.9 % accuracy for late deceleration and 98.9 % for prolonged deceleration. The proposed multimodal deep learning algorithm serves as a reliable decision support tool for clinicians, significantly improving the detection and assessment of specific FHR events, which are crucial for fetal health monitoring.

Improving Image Quality and Diagnostic Performance of CCTA in Patients with Challenging Heart Rate Conditions using a Deep Learning-based Motion Correction Algorithm

Challenging HR conditions, such as elevated Heart Rate (HR) and Heart Rate Variability (HRV), are major contributors to motion artifacts in Coronary Computed Tomography Angiography (CCTA). This study aims to assess the impact of a deep learning-based motion correction algorithm (MCA) on motion artifacts in patients with challenging HR conditions, focusing on image quality and diagnostic performance of CCTA. This retrospective study included 240 patients (mean HR: 88.1 ± 14.5 bpm; mean HRV: 32.6 ± 45.5 bpm) who underwent CCTA between June, 2020 and December, 2020. CCTA images were reconstructed with and without the MCA. The signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were measured to assess objective image quality. Subjective image quality was evaluated by two radiologists using a 5-point scale regarding vessel visualization, diagnostic confidence, and overall image quality. Moreover, all vessels with scores ≥ 3 were considered clinically interpretable. The diagnostic performance of CCTA with and without MCA for detecting significant stenosis (≥ 50%) was assessed in 34 patients at both per-vessel and per-patient levels, using invasive coronary angiography as the reference standard. The MCA significantly improved subjective image quality, increasing the vessel interpretability from 89.9% (CI: 0.88-0.92) to 98.8% (CI: 0.98-0.99) (p < 0.001). The use of MCA resulted in significantly higher diagnostic performance in both patient-based (AUC: 0.83 vs. 0.58, p = 0.04) and vessel-based (AUC: 0.92 vs. 0.81, p < 0.001) analyses, with the vessel-based accuracy notably increased from 79.4% (CI: 0.72-0.86) to 91.2% (CI: 0.85-0.95) (p = 0.01). There were no significant differences in objective image quality between the two reconstructions. The mean effective dose in this study was 2.8 ± 1.1 mSv. The use of MCA allows for obtaining high-quality CCTA images and superior diagnostic performance with low radiation exposure in patients with elevated HR and HRV.

Development of a deep learning-based auto-segmentation algorithm for peritoneal metastases using CT image analysis of ovarian cancer

Development of a deep learning-based auto-segmentation algorithm for peritoneal metastases using CT image analysis of ovarian cancer

Simulating the Cellular Context in Synthetic Datasets for Cryo-Electron Tomography.

Cryo-electron tomography (cryo-ET) allows to visualize the cellular context at macromolecular level. To date, the impossibility of obtaining a reliable ground truth is limiting the application of deep learning-based image processing algorithms in this field. As a consequence, there is a growing demand of realistic synthetic datasets for training deep learning algorithms. In addition, besides assisting the acquisition and interpretation of experimental data, synthetic tomograms are used as reference models for cellular organization analysis from cellular tomograms. Current simulators in cryo-ET focus on reproducing distortions from image acquisition and tomogram reconstruction, however, they can not generate many of the low order features present in cellular tomograms. Here we propose several geometric and organization models to simulate low order cellular structures imaged by cryo-ET. Specifically, clusters of any known cytosolic or membrane-bound macromolecules, membranes with different geometries as well as different filamentous structures such as microtubules or actin-like networks. Moreover, we use parametrizable stochastic models to generate a high diversity of geometries and organizations to simulate representative and generalized datasets, including very crowded environments like those observed in native cells. These models have been implemented in a multiplatform open-source Python package, including scripts to generate cryo-tomograms with adjustable sizes and resolutions. In addition, these scripts provide also distortion-free density maps besides the ground truth in different file formats for efficient access and advanced visualization. We show that such a realistic synthetic dataset can be readily used to train generalizable deep learning algorithms.

Deep learning-based segmentation of kidneys and renal cysts on T2-weighted MRI from patients with autosomal dominant polycystic kidney disease

BackgroundOur aim was to train and test a deep learning-based algorithm for automatically segmenting kidneys and renal cysts in patients with autosomal dominant polycystic kidney disease (ADPKD).MethodsWe retrospectively selected all ADPKD patients who underwent renal MRI with coronal T2-weighted imaging at our institution from 2008 to 2022. The 20 most recent examinations constituted the test dataset, to mimic pseudoprospective enrolment. The remaining ones constituted the training dataset to which eight normal renal MRIs were added. Kidneys and cysts ground truth segmentations were performed on coronal T2-weighted images by a junior radiologist supervised by an experienced radiologist. Kidneys and cysts of the 20 test MRIs were segmented by the algorithm and three independent human raters. Segmentations were compared using overlap metrics. The total kidney volume (TKV), total cystic volume (TCV), and cystic index (TCV divided by TKV) were compared using Bland–Altman analysis.ResultsWe included 164 ADPKD patients. Dice similarity coefficients ranged from 85.9% to 87.4% between the algorithms and the raters’ segmentations and from 84.2% to 86.2% across raters’ segmentations. For TCV assessment, the biases ± standard deviations (SD) were 3–19 ± 137–151 mL between the algorithm and the raters, and 22–45 ± 49–57 mL across raters. The algorithm underestimated TKV and TCV in two outliers with TCV > 2800 mL. For cystic index assessment, the biases ± SD were 2.5–6.9% ± 6.7–8.3% between the algorithm and the raters, and 2.1–9.4 ± 7.4–11.6% across raters.ConclusionThe algorithm’s performance fell within the range of inter-rater variability, but large TKV and TCV were underestimated.Relevance statementAccurate automated segmentation of the renal cysts will enable the large-scale evaluation of the prognostic value of TCV and cystic index in ADPKD patients. If these biomarkers are prognostic, then automated segmentation will facilitate their use in daily routine.Key PointsCystic volume is an emerging biomarker in ADPKD.The algorithm’s performance in segmenting kidneys and cysts fell within interrater variability.The segmentation of very large cysts, under-represented in the training dataset, needs improvement.Graphical

Intelligent Control Framework for Improving Energy System Stability Through Deep Learning-Based Modal Optimization Scheme

Ensuring the stability of power systems is essential to promote energy sustainability. The integrated operation of these systems is critical in sustaining modern societies and economies, responding to the increasing demand for electricity and curbing environmental consequences. This study focuses on the optimization of energy system stability through the coordination of power system stabilizers (PSSs) and power oscillation dampers (PODs) in a single-machine infinite bus energy grid configuration that has flexible AC alternating current transmission system (FACTS) devices. Intelligent control strategies using PSS and POD techniques are suggested to increase power system stability and generate supplementary control signals for both the generator excitation system and FACTS device switching control. An intelligent optimal modal control framework equipped with deep learning methods is introduced to control the generator excitation system and thyristor-controlled series capacitor (TCSC). By optimally choosing the weighting matrix Q and implementing close-loop pole shifting, an optimal modal control approach is formulated. To harness its adaptive potential in fine-tuning controller parameters, an auxiliary deep learning-based optimization algorithm with actor–critic architecture is implemented. This comprehensive technique provides a promising path to effectively reduce electromechanical oscillations, thereby enhancing voltage regulation and transient stability in power systems.

Evaluation of a Vendor-Agnostic Deep Learning Model for Noise Reduction and Image Quality Improvement in Dental CBCT.

To assess the impact of a vendor-agnostic deep learning model (DLM) on image quality parameters and noise reduction in dental cone-beam computed tomography (CBCT) reconstructions. This retrospective study was conducted on CBCT scans of 93 patients (41 males and 52 females, mean age 41.2 years, SD 15.8 years) from a single center using the inclusion criteria of standard radiation dose protocol images. Objective and subjective image quality was assessed in three predefined landmarks through contrast-to-noise ratio (CNR) measurements and visual assessment using a 5-point scale by three experienced readers. The inter-reader reliability and repeatability were calculated. Eighty patients (30 males and 50 females; mean age 41.5 years, SD 15.94 years) were included in this study. The CNR in DLM reconstructions was significantly greater than in native reconstructions, and the mean CNR in regions of interest 1-3 (ROI1-3) in DLM images was 11.12 ± 9.29, while in the case of native reconstructions, it was 7.64 ± 4.33 (p < 0.001). The noise level in native reconstructions was significantly higher than in the DLM reconstructions, and the mean noise level in ROI1-3 in native images was 45.83 ± 25.89, while in the case of DLM reconstructions, it was 35.61 ± 24.28 (p < 0.05). Subjective image quality assessment revealed no statistically significant differences between native and DLM reconstructions. The use of deep learning-based image reconstruction algorithms for CBCT imaging of the oral cavity can improve image quality by enhancing the CNR and lowering the noise.

Deep learning-based algorithm for staging secondary caries in bitewings.

Despite the notable progress in developing artificial intelligence (AI)-based tools for caries detection in bitewings, limited research has addressed the detection and staging of secondary caries. Therefore, we aimed to develop a Convolutional neural network (CNN)-based algorithm for these purposes using a novel approach for determining lesion severity. We used a dataset from a Dutch dental practice-based research network containing 2,612 restored teeth in 413 bitewings from 383 patients aged 15 to 88 years and trained the Mask R-CNN architecture with a Swin Transformer backbone. Two-stage training fine-tuned caries detection accuracy and severity assessment. Annotations of caries around restorations were made by two evaluators and checked by two other experts. Aggregated accuracy metrics (mean ± Standard Deviation - SD) in detecting teeth with secondary caries were calculated considering two thresholds: detecting all lesions and dentine lesions. The correlation between the lesion severity scores obtained with the algorithm and the annotators' consensus was determined using the Pearson correlation coefficient and Bland-Altman plots. Our refined algorithm showed high specificity in detecting all lesions (0.966 ± 0.025) and dentine lesions (0.964 ± 0.019). Sensitivity values were lower: 0.737 ± 0.079 for all lesions and 0.808 ± 0.083 for dentine lesions. The areas under ROC curves (SD) were 0.940 (0.025) for all lesions and 0.946 (0.023) for dentine lesions. The correlation coefficient for severity scores was 0.802. We developed an improved algorithm to support clinicians in detecting and staging secondary caries in bitewing, incorporating an innovative approach for annotation, considering the lesion severity as a continuous outcome.