Portraits and memory: Investigating HF-rTMS effects on episodic recall through a sham-controlled fMRI study.

To evaluate the influence of driver power and slice thickness on magnetic resonance elastography (MRE)-derived liver stiffness in human subjects and a tissue-mimicking phantom. A prospective single-center observational study was performed involving 34 adult participants (20 males; median age: 29.0 ± 29.3 years) and a standard phantom. MRE was conducted on a 1.5T scanner using a 2D gradient-recalled echo-based sequence. Human data were acquired at driver powers of 25%-100% and slice thicknesses of 5 and 10 mm. The phantom was scanned with driver powers of 10%-100% and slice thicknesses from 2 to 10 mm. Two region-of-interest (ROI) approaches were applied: (A) fixed-location ROIs and (B) largest measurable ROIs. Stiffness maps were generated using the MMDI inversion algorithm. Statistical analysis included repeated measures ANOVA, paired t-tests or Wilcoxon signed-rank tests, and intraclass correlation coefficients (ICCs). Measured whole-liver stiffness varied modestly over the driver power range, with the effect reaching statistical significance in Method B analysis (p = 0.037). Measurements of individual slices varied more significantly with driver power (Method A: p = 0.031; Method B: p < 0.001). Slice thickness had a significant effect on individual slice measurements (Method B: p = 0.007) but not on whole-liver measurements. In phantom studies, both parameters significantly influenced stiffness (p < 0.001 to p = 0.017). ICCs indicated good to excellent repeatability in vivo (0.79-0.96) and moderate to good repeatability in the phantom (0.62-0.89). The results demonstrate that measured MRE-derived liver stiffness can be influenced by driver power and slice thickness, underscoring the need for standardization of these parameters in both clinical practice and research.

Early differential diagnosis of bipolar disorder (BD) and unipolar depression (UD) remains a major clinical challenge, especially during the initial depressive episode. This study aimed to investigate whether cerebellar gray matter volume could serve as a potential neuroimaging biomarker to distinguish BD from UD at an early stage. Structural MRI data were obtained from 42 patients with BD, 48 patients with UD, and 83 matched healthy controls. The cerebellum was parcellated into 28 lobules and 7 functional networks using the SUIT anatomical and Buckner-Yeo functional atlases. Voxel-based and region-of-interest (ROI) analyses were conducted, and a machine learning model was applied. Whole-cerebellum analyses revealed that the total cerebellar gray matter volume in UD patients was significantly larger than in BD patients. Voxel-based studies indicated that the regions of Vermis IX, Vermis VI, and Left VI were significantly smaller in the BD group compared to the UD group. ROI-based comparisons showed a significant higher in the volume of gray matter in the limbic network in UD patients compared to BD. Furthermore, the gray matter volume of the limbic network in the dorsal attention network and the default mode network was significantly reduction in BD patients compared to healthy controls. The machine learning model constructed using cerebellar lobules with significant inter-group differences achieved an accuracy of 76.3% and a sensitivity of 83.0%. Cerebellar gray matter volume may serve as a clinical marker to differentiate the depressive phase of bipolar disorder (after the first manic or hypomanic episode) from unipolar depression.

Mass spectrometry imaging (MSI) is a powerful tool for mapping the spatial distribution of biomolecules in biological samples. Among the various MSI techniques, nanospray desorption electrospray ionization (nano-DESI) is ideally suited for quantitative imaging of a wide range of biomolecules in biological tissues due to its capabilities as an ambient, liquid extraction-based technique. In this study, we used nano-DESI MSI to investigate the effects of Scn2a gene deficiency in the mouse brain. Scn2a, which encodes the voltage-gated sodium channel NaV1.2, is critical to neuronal excitability, and its dysfunction is linked to epilepsy and neurodevelopmental disorders such as autism. Despite its importance, the molecular alterations associated with Scn2a dysfunction are still poorly understood. Herein, we present the first comprehensive study of regional lipid and metabolite alterations associated with Scn2a deficiency, achieved by comparing brain tissues from wild-type (WT) and Scn2a homozygous gene-trap (HOM) mice. Nano-DESI MSI experiments were performed on an Orbitrap mass spectrometer in both positive and negative ionization modes, with three biological replicates per group to ensure reproducible detection and broad coverage of biomolecules. Region-of-interest (ROI) analysis revealed multiple species with altered abundance in the HOM mouse brain. Notably, several phosphatidylethanolamine (PE) lipids were observed at higher abundance in different regions of the brain. For example, PE(O-36:5) is more abundant in both the cortex and hippocampus of the HOM brains, while PE(40:4) is more abundant in the hippocampus. Meanwhile, several lipid species, including phosphatidylserine, PS(38:1), were at lower abundance in the cortex. In contrast, abundant structural lipids, including phosphatidylinositol, PI(38:4), and phosphatidylcholine, PC(34:1), showed no significant differences between WT and HOM brains. Our findings offer new insights into the lipid alterations underlying epilepsy and related neurodevelopmental disorders associated with Scn2a deficiency.

To investigate the potential of the unidirectional cellular water efflux rate constant (kio) from DCE-MRI data in prostate MR imaging. High-temporal-resolution prostate DCE-MRI data were modeled using both the fast-exchange-limit (FXL) Tofts' model as well as the water-exchange-sensitized shutter-speed model (SSM). In the SSM, kio was included as an additional fitting parameter. Lesion and normal-appearing (NA) prostate tissue region-of-interest (ROI) data were analyzed and categorized into FXL or non-FXL conditions based on results from the two models. A global upper limit of kio detectable by prostate DCE-MRI with the SSM was presented. While many lesion voxels exhibited sensitivity to kio with the SSM, a substantial portion remained in the FXL condition despite greater contrast agent extravasation than in NA tissue. The fraction of FXL voxels was higher in lesions than in NA tissue. Applying a global detectable kio upper limit increased the difference between lesion and NA ROIs, improving lesion characterization. SSM-derived FXL and non-FXL contrasts may serve as novel imaging biomarkers for prostate cancer surveillance. Advances in MRI technology and more potent contrast agents are expected to enhance the accuracy of kio quantification, potentially enabling its integration into clinical mpMRI.

Recently, a posterior pathway for fluid drainage from the retina to the meningeal lymphatics in the optic nerve (ON) sheath was identified in rodents using intravitreal imaging tracers directly injected into the ocular-globe. Fluid and solute clearance along this pathway may be associated with many diseases. However, intravitreal tracers are rarely used in clinical imaging. As intravenous Gadolinium-based-contrast-agent (GBCA) can enter the globe via the blood-ocular-barriers, it may provide an alternative approach to image this pathway. To establish a clinically feasible intravenous GBCA-based MRI approach for tracking fluid and solute transport along the posterior lymphatic pathway in the ocular glymphatic system. This prospective study was conducted from March 2021 to September 2022 in healthy participants. Dynamic-susceptibility-contrast-in-the-CSF (cDSC) MRI was performed before, immediately and 4 hours after intravenous-GBCA administration to track GBCA distribution in aqueous humor (AH) and cerebrospinal fluid (CSF) in regions-of-interest (ROIs) in the globe (anterior-cavity, vitreous-body), in the intraorbital and extraorbital ON, and in the intracranial CSF space proximal to the ON (chiasmatic-cistern, interpeduncular-cistern). Kruskal-Wallis tests with post-hoc Dunn's tests were used for group comparisons. Sixteen healthy participants (mean age±SD: 51±21 years, 5 men) were recruited. Intravenous-GBCA enhancement was observed in all ROIs immediately after injection. At 4-hour-post-GBCA, the vitreous body showed a trend of smaller enhancement area (55±11% versus 49±11%, P =.14) and lower GBCA-concentration (0.044±0.014 versus 0.028±0.010 mmol/L, P =.07) compared to immediate-post-GBCA. The intraorbital ON showed more widespread enhancement (39±5% versus 59±6%, P =.01) and significantly higher GBCA-concentration (0.023±0.009 versus 0.059±0.015 mmol/L, P <.001) at 4-hour-post-GBCA. Dynamic fluid and solute transportation along the posterior lymphatic pathway in the ocular glymphatic system in healthy participants was measured by tracking intravenous-GBCAs entering the globe via the blood-ocular-barriers using cDSC-MRI. Dynamic fluid and solute transportation along the posterior lymphatic pathway in the ocular glymphatic system in healthy participants was measured by tracking intravenous-Gadolinium-based-contrast-agents entering ocular-globe via blood-ocular-barriers using dynamic-susceptibility-contrast-in-the-CSF (cDSC)-MRI. In this prospective study of sixteen participants, Gadolinium-based-contrast-agent (GBCA)-induced signal changes were detected in the aqueous-humor and cerebrospinal fluid immediately following intravenous administration using dynamic-susceptibility-contrast-in-the-CSF (cDSC)-MRI. At 4-hour-post-GBCA, the vitreous-body showed a trend of smaller enhancement area (55±11% versus 49±11%, P =.14) and lower GBCA-concentration (0.044±0.014 versus 0.028±0.010 mmol/L, P =.07) compared to immediate-post-GBCA. The intraorbital-optic-nerve showed more widespread enhancement (39±5% versus 59±6%, P =.01) and higher GBCA-concentration (0.023±0.009 versus 0.059±0.015 mmol/L, P <.001) at 4-hour-post-GBCA.

This study aimed to evaluate the feasibility and accuracy of automated contrast-to-noise ratio (CNR) analysis in chest CT using the open-source body and organ analysis (BOA) framework and to validate segmentation modifications for reproducible image-quality assessment. This retrospective study analyzed 100 contrast-enhanced chest CTs (mean age 60.2 ± 15 years; 40% female; 50 CTA, 50 CTPA) and validated the approach in an external cancer imaging archive (TCIA) cohort (n = 100). Automated BOA segmentations of the aorta, pulmonary trunk, and paraspinal muscles were modified by fat subtraction and binary erosion and compared with manual measurements from three radiologists. Agreement was assessed using statistical testing, Bland-Altman analysis, and intraclass correlation coefficients (ICC). Unmodified BOA segmentations yielded significantly lower CNRs than manual measurements (all p < 0.01, mean difference up to 6.3). Fat subtraction and binary erosion progressively reduced deviations, with the optimized variant (m_erode6 combined with a_erode6 or p_erode6) showing no significant differences from radiologists (p ≥ 0.35). In the external TCIA validation cohort (n = 100), agreement was excellent (ICC 0.89-0.93), and Bland-Altman analysis demonstrated minimal bias (Aorta: 0.16 [limits of agreement (LoA) -3.47 to 3.80]; PT: 0.42 [LoA -4.03 to 4.87]). A minimally modified open-source segmentation framework enables fully automated, reproducible CNR assessment in chest CT, achieving expert-level agreement, including robust performance in external validation. This scalable alternative to manual region-of-interest (ROI) measurement streamlines image-quality assessment, facilitates protocol optimization, and provides standardized metrics ready for integration into AI workflows. This study provides a validated, fully automated method for quantitative CT image quality assessment, reducing observer dependence and enabling consistent evaluation across scanners, protocols, and institutions, thereby supporting reproducible image quality metrics in clinical routine. Automated CNR assessment enables objective and reproducible evaluation of image quality in CTA and CTPA. Adjustments of the segmentation strategy can substantially improve the accuracy of automated measurements. The fully automated approach provides a foundation for standardized and scalable CT image quality analysis in research and clinical practice.

To evaluate the in vivo efficacy of bone morphogenetic protein 9 (BMP9) relative to bone morphogenetic protein 2 (BMP2) in a beagle peri-implant critical-size defect model. Peri-implant defects were created in beagle dogs and treated with either a collagen sponge (CS) or deproteinized bovine bone mineral with collagen (OCS-B Xenomatrix collagen [OCS-BC]), with or without BMP2 or BMP9. After 8 weeks, bone regeneration and osseointegration were evaluated using radiographic, histological, and biomechanical analyses. Radiographic analysis demonstrated that BMP treatment significantly increased peri-implant mineralization, with bone mineral density increasing from 0.597±0.151 (control) to 0.896±0.173 in the CS+BMP9 group (P<0.0001). Histological and histomorphometric analyses corroborated these findings, showing greater new bone formation and higher bone-to-implant contact (BIC) (7.943±7.048 vs. 68.90±20.27 in the CS+BMP9 group; P<0.05) without overt inflammatory reactions. Region-of-interest (ROI)-based micro-computed tomography analysis (200 μm) and resonance frequency analysis further supported improved osseointegration and stability, with osseointegration increasing from 20.32±6.976 to 70.37±0.785 in the CS+BMP9 group (P<0.0001) and the implant stability quotient (ISQ) peaking at 70.92±3.523 in the OCS-BC+BMP9 group (P<0.0001). Two-way analysis of variance indicated scaffold-dependent magnitudes for bulk endpoints, whereas interaction terms were not significant for BIC, ROI-based osseointegration, or ISQ, supporting preservation of BMP9-associated benefits across scaffold types for these outcomes. BMP9 produced overall improvements comparable to BMP2 and may provide more reproducible osseointegration-related outcomes across scaffold types.

Quantitative phenotyping is essential to studies of animal behavior, enabling systematic analysis of variation arising from natural diversity or experimental manipulation. High-throughput behavioral assays that can simultaneously test multiple animals support sufficiently powered studies of behavioral variation, but accurate tracking of each animal is critical. Furthermore, behavioral tasks and experimental arenas span a wide range of complexity, from the reaction of a single larval zebrafish to an acoustic stimulus to associative conditioning in cue-rich environments. Here, we developed and validated StrIPETrack (Structural similarity-based Image Processing for Estimation and Tracking), a Python-based, modular animal tracking software designed for flexible region-of-interest (ROI) definitions and extensibility across assays. We show that StrIPETrack measures activity comparably to our previous LabVIEW-based zebrafish tracking software and detects similar behavioral differences between wild-type clutches. In addition, StrIPETrack accurately captures behavior in a complex arena: the Y-maze. Our approach for analyzing Y-maze navigation yields an expanded set of metrics beyond turn count and direction, revealing more subtle behavioral variation. Overall, this versatile software can be applied to monitor the activity of multiple animals in parallel in both simple, high-throughput and more complex assays, and it can be readily adapted to new paradigms.

Magnetic resonance elastography (MRE) has become an essential noninvasive tool for quantitative assessment of liver stiffness and fibrosis staging, particularly in patients with steatotic liver disease. To enhance standardization and reproducibility, the Quantitative Imaging Biomarkers Alliance (QIBA) released an updated liver MRE profile in 2023, replacing the previous 2020 version. However, the quantitative continuity between these two protocol versions, particularly for longitudinal liver stiffness measurements, has not been sufficiently validated. This study aimed to evaluate the quantitative equivalence and clinical utility of a QIBA-2023-compliant MRE protocol through direct comparison with the conventional QIBA-2020-compliant protocol in the same subjects, with emphasis on longitudinal liver stiffness assessment and image quality. This retrospective study included 130 consecutive patients who underwent liver MRE in 2024. Each patient was examined using both the conventional QIBA-2020-compliant protocol and the updated QIBA-2023-compliant protocol during a single imaging session. Liver stiffness values were compared using Spearman correlation, Bland-Altman analysis, and intraclass correlation coefficients (ICC). Measurement variability and repeatability were evaluated with reference to QIBA-defined repeatability metrics. Image quality was evaluated by comparing region-of-interest (ROI) size and measurable liver parenchymal area excluding low-confidence (cross-hatched) regions. Mean liver stiffness values were 3.62 ± 1.81kPa for the conventional protocol and 3.69 ± 1.94kPa for the updated protocol, with a very strong correlation (r = 0.98, p < 0.001) and excellent agreement (ICC = 0.988). Bland-Altman analysis showed a minimal mean bias of 0.07kPa with 95% limits of agreement from - 0.48 to 0.63kPa, which were within established MRE repeatability thresholds. Although the difference between protocols reached statistical significance, the absolute difference was not clinically meaningful. The updated protocol produced significantly larger ROIs and a greater measurable liver parenchymal area, consistent with reduced imaging artifacts and improved image quality. The QIBA-2023-compliant MRE protocol demonstrates quantitative equivalence to the conventional QIBA-2020 protocol while offering improved artifact suppression and expanded measurable liver parenchyma. These findings support use of the updated protocol for reliable longitudinal liver stiffness assessment without compromising diagnostic performance.

Older adults often exhibit reduced inhibitory control accompanied by altered recruitment of prefrontal networks. Whether the emotional context changes these age-related neural patterns during response inhibition remains unclear. In this study, 31 older adults and 19 young adults completed four blocks of a Go/No-Go paradigm while bilateral prefrontal cortex (PFC) activity was recorded using functional near-infrared spectroscopy (fNIRS). Blocks 1 and 2 comprised a neutral (non-emotional) Go/No-Go task using geometric shapes, while blocks 3 and 4 comprised an emotional Go/No-Go task using facial expressions (happy or angry as Go; neutral as No-Go). Task-evoked oxygenated (HbO) and deoxygenated hemoglobin (HbR) responses were quantified and analyzed at both the region-of-interest (ROI) and channel levels using linear mixed-effects models. Behaviorally, older adults showed markedly lower No-Go accuracy than young adults (p < 0.001), and emotional blocks further reduced the accuracy in both groups (p < 0.001). Crucially, reaction time analyses revealed a significant group × condition interaction (p < 0.001): young adults exhibited strategic slowing in the emotional condition, whereas older adults failed to modulate their response speed. Neurally, ROI analyses revealed robust main effects of group (older > young) and condition (emotional < neutral) on HbO across dorsolateral, dorsomedial, and ventromedial ROIs after false discovery rate correction, whereas group × condition interactions were not significant. Brain-behavior analyses revealed that higher prefrontal activation in older adults significantly predicted a poorer performance, supporting a neural inefficiency account. Translationally, these findings suggest that portable fNIRS measures of PFC inefficiency may serve as a scalable biomarker to identify older adults at risk for inhibitory-control failures-especially when emotion is present-and to track neural targets and treatment response in cognitive or emotion-regulation interventions.

Objective: The classic excitation/inhibition dichotomy may be insufficient to describe rTMS mechanisms in the aging brain. This study investigated immediate whole-brain resting-state functional connectivity effects of 10 Hz (high-frequency) and 1 Hz (low-frequency) rTMS over the left dorsolateral prefrontal cortex (DLPFC) in healthy older adults. Methods: Thirty healthy older adults (aged 60-75 years) participated in a randomized, single-blind, crossover study, and underwent 20-min 10 Hz and 1 Hz rTMS in separate visits. A 106-channel fNIRS system was used to record resting-state activity before and immediately after each intervention. Functional connectivity was analyzed at the channel, region-of-interest (ROI) and network summary levels, including graph-theoretic metrics and distance-stratified connectivity summaries. Results: At the network summary level, 10 Hz stimulation was associated with relatively more positive changes in global topology and spatially distributed connectivity summaries, whereas 1 Hz stimulation showed the opposite overall trend. In the graph-theoretic analyses, stimulation frequency × time interaction effects were observed for global efficiency, local efficiency, clustering coefficient, and mean node strength. At the edge level, only a small number of effects survived FDR correction, and the broader connection-wise patterns were therefore interpreted as exploratory. Uncorrected analyses suggested widespread enhancement after 10 Hz stimulation and widespread reduction after 1 Hz stimulation, together with localized paradoxical effects, including selective decreases after 10 Hz and selective increases after 1 Hz (e.g., bilateral primary motor cortex connectivity). Conclusions: These findings suggest that 10 Hz and 1 Hz rTMS over the left DLPFC are associated with different patterns of immediate whole-brain network reconfiguration in healthy older adults. The presence of localized paradoxical effects further suggests that rTMS responses in the aging brain may involve more complex forms of reorganization than a simple excitatory/inhibitory dichotomy would predict. Significance: The present study provides preliminary support for a network-level perspective on neuromodulation in older adults and highlights the value of whole-brain fNIRS for characterizing distributed responses to rTMS. Larger, sham-controlled, behavior-linked, and longitudinal studies are needed to determine the robustness and functional significance of these effects.

Prediction of soybean fatty-acid composition from hyperspectral imaging with spectral feature processing and structured tabular modeling.

Abstract Chemical species tomography (CST) is a powerful technique for non-intrusive reconstruction of temperature and species concentration fields in reacting flows. However, conventional full-domain reconstruction methods often suffer from artifacts near the flame boundaries, especially when the absorbing region is confined within a smaller zone. To address this limitation, we propose an absorption-guided region-of-interest (ROI) refinement approach for CST. The method begins with a full-field reconstruction to estimate absorption densities, followed by threshold-based extraction of the absorption-meaningful ROI. A second reconstruction is then performed within this refined ROI while representing the ambient region with a single unknown absorption value. This strategy reduces artifacts at ROI boundaries and improves reconstruction accuracy. The method is firstly validated using synthetic data, demonstrating a 20.90% reduction in average relative squared error for temperature and an 8.90% reduction for mole fraction compared to the adoption of conventional ROI. Experiments further shows improved temperature and water vapor concentration reconstructions on multi-burner flame configurations.

ABSTRACT Open Radio Access Network (O‐RAN) fronthaul links face stringent bandwidth, latency, and computational constraints, which become particularly critical when transmitting high‐resolution medical images. This paper proposes an adaptive region‐aware image compression framework for healthcare imaging over O‐RAN that reduces fronthaul load while preserving diagnostically relevant information. Each image is partitioned into Region‐of‐Interest (ROI) and Non‐ROI areas and compressed using independent quantisation parameters. An optimisation model is formulated to minimise transmitted data size subject to ROI and Non‐ROI quality constraints, end‐to‐end latency bounds and computational limits at O‐RAN nodes. The framework is evaluated using two medical imaging datasets (chest X‐rays and bone fracture X‐rays), where empirical rate–distortion and quality models are derived and validated. Results demonstrate substantial fronthaul bandwidth reduction—achieving compression ratios up to 416:1—while maintaining ROI quality and diagnostic accuracy above 97%. These findings highlight the effectiveness of region‐aware optimisation for bandwidth‐efficient healthcare imaging in O‐RAN environments.

Quantitative immunofluorescence is widely used to assess molecular expression and cellular distribution across biological tissues, yet the analysis of large image datasets remains time-consuming and prone to user-dependent variability. To address these limitations, we herein developed a semi-automated workflow that integrates ImageJ/Fiji for image processing, StarDist for nuclear segmentation, and spreadsheet- or Python-based routines for data curation. The pipeline standardizes critical analytical steps, including scale calibration, region-of-interest (ROI) definition, channel selection, and z-stack handling, while preserving essential metadata through a structured file-naming system. Optical density and cell-number metrics are exported automatically in a consistent format, enabling efficient consolidation into a unified dataset. Subsequent curation can be performed either manually in a spreadsheet software or fully automatically through custom Python scripts, allowing extraction of sample identifiers, regions, and markers, as well as calculation of normalized intensity values. Comparison with existing protocols proved that this workflow adheres to widely accepted quantification principles while markedly improving reproducibility, consistency, and analytical throughput. This method offers a straightforward, transparent, and scalable solution for fluorescence-based quantification suitable for laboratories with varying levels of computational expertise.

To compare two region-of-interest (ROI) measurement methods based on mDIXON-Quant for distinguishing healthy volunteers (HV), chronic kidney disease (CKD) patients with low (≤ 15%) and high (> 15%) glomerulosclerosis (GS), and establish a non-invasive auxiliary protocol. A total of 42 CKD patients and 25 HV underwent mDIXON-Quant MRI. CKD patients were stratified into low (n = 24) and high (n = 18) GS groups via renal biopsy. Two ROI methods were applied: Method 1 (cortex/medulla separation, 15–20 mm² circular ROIs at renal hilum, upper/lower poles) and Method 2 (whole-kidney contour tracing). R2* and fat fraction (FF) parameters were analyzed using ANOVA, LSD tests, and DeLong tests for diagnostic efficacy (AUC). R2* parameters (RcR2*, RR2*) showed significant differences across groups (P < 0.05), increasing with GS severity. LcR2* and RmR2* distinguished HV from high GS (P < 0.05), while LFF differentiated low vs. high GS (P < 0.05). Combining FF with R2* improved AUC from 0.685 to 0.877 (P < 0.05) for low/high GS discrimination. Method 1’s RcR2* (AUC = 0.957) and Method 2’s RR2* (AUC = 0.9) were most accurate for distinguishing high GS from HV, with no significant difference between them (P > 0.05). RcR2* outperformed medullary parameters (P < 0.05). The present study demonstrates that employing the Method 2(whole-kidney contour tracing) based on mDIXON-Quant technology enables auxiliary quantitative assessment of the severity of GS via the R2 parameter of the right kidney, thereby confirming its role as an adjunctive tool in GS evaluation.

Background/Objectives: This study aimed to automatically classify Zirconia-based fixed restorations and porcelain-fused-to-metal (PFM) restorations on panoramic radiographs using an artificial intelligence-based model. Unlike previous studies that mainly focused on classifying types of restorations (e.g., crowns, fillings, implants), this research concentrated on material-based differentiation, aiming to provide a more specific contribution to clinical decision support systems. Method: Panoramic radiographs obtained from the archive of Süleyman Demirel University Faculty of Dentistry were included in this study. Radiographs with poor image quality or insufficient visibility of the restoration area were excluded. A total of 593 cropped region-of-interest (ROI) images, labeled by expert prosthodontists using ImageJ software (version 1.54r; National Institutes of Health, Bethesda, MD, USA), were included in the analysis. In order to reduce class imbalance, data augmentation was applied only for images in the Zirconia-based fixed restorations class. By using various image processing techniques such as rotation, reflection and brightness change, the number of samples in the zirconia-based restorations class was increased and thus a balanced dataset was obtained with a close number of samples for both classes. For model training, the pre-trained VGG16 architecture was used with a transfer learning method, and the final layers were retrained and fine-tuned. The model was configured specifically for binary classification. The entire dataset was randomly split into 70% training, 20% validation, and 10% testing. Model performance was evaluated using accuracy, F1-score, sensitivity, and specificity. Results: The model correctly classified 90 out of 94 images in the test dataset, achieving an overall accuracy rate of 96%. For both classes, the precision, recall, and F1-score values were measured in the range of 95% to 96%. Additionally, the Area Under the Curve (AUC) of the ROC curve was calculated as 0.994, and the Average Precision (AP) score was determined to be 0.995. According to the confusion matrix results, only 4 images were misclassified, consisting of 2 false positives and 2 false negatives. Conclusions: The deep learning model demonstrated high accuracy in differentiating zirconia and metal-supported porcelain restorations on panoramic radiographs, suggesting that material-based AI classification may support clinical decision-making in restorative dentistry.

This study proposes an advanced short-term Direct Normal Irradiance (DNI) prediction model for Concentrated Solar Power (CSP) systems, integrating a convolutional neural network (CNN) with a fitted clear-sky DNI model. Leveraging all-sky images and historical DNI data, the model precisely identifies cloud motion patterns through dense optical flow analysis and forecasts DNI using a targeted region-of-interest (ROI) approach. When maximum cloud pixel velocity falls below 5 pixels per minute, the clear-sky DNI model or persistence model directly applies; for higher-velocity conditions, the CNN predicts the clear-sky index to dynamically adjust the forecast. Experimental validation across diverse weather conditions demonstrates superior accuracy, achieving significantly lower normalized Mean Absolute Errors (nMAEs) and normalized Root Mean Squared Errors (nRMSEs) for various forecast horizons under cloudy skies compared to recent state-of-the-art deep learning approaches. This work delivers a robust solution for preventing thermal shock in the receiver and improving the CSP operational stability.

Manual single-slice lumbar quantitative computed tomography (QCT) depends on operator-driven slice selection and trabecular region-of-interest (ROI) placement. We developed a fully automated single-slice workflow for vertebral trabecular Hounsfield unit (HU) measurement that combines unsuitable-slice prescreening, dual-purpose segmentation, intra-patient slice-quality ranking, and a deterministic inner ROI rule. The pipeline includes an Eligibility Gate, QC-Envelope segmentation for broad, vertebral- and usability-preserving delineation, PairRank-Swin for best-slice selection, and dedicated trabecular segmentation for final quantitative analysis. In the independent external cohort, 4 cases were considered non-evaluable by both manual review and the pipeline, and 2 additional borderline-quality cases were manually measured but rejected by the pipeline; therefore, paired HU agreement analysis included 44 evaluable cases. Agreement remained high, with Pearson’s r = 0.987, Lin’s CCC = 0.985, mean bias −0.44 HU, and limits of agreement from −14.88 to +13.99 HU. Coverage was 84.1% within ±10 HU and 97.7% within ±15 HU. Ablation analysis showed that slice ranking and ROI erosion were the most critical components. In an open module-level baseline comparison, QC-Envelope segmentation substantially outperformed TotalSegmentator. This workflow provides high agreement with expert HU measurement while preserving reviewable intermediate outputs.