Search Results (477)

Background subtraction (BS) is a fundamental technique in intelligent transport vision systems, widely used to detect and track moving objects, such as vehicles, pedestrians and obstacles, in driving environments. It plays a crucial role in advanced driver-assistance systems (ADAS) and autonomous driving by enabling scene understanding and real-time motion analysis. However, BS processing must be optimized when targeting real-time processing on resource-constrained embedded systems, which present significant challenges due to limited computational power, memory constraints, and strict real-time requirements. Among the most commonly used BS techniques, the Codebook model and Gaussian Mixture Models (GMM) are known for their higher accuracy and light-model compared to many deep learning-based BS. In this work, we propose a fully heterogeneous CPU and GPU parallel implementation of both Codebook and GMM algorithms with an auto-load balancing over the processing units. This approach has been evaluated on the low-cost Jetson Orin Nano platform from NVIDIA, enabling efficient workload balancing across heterogeneous hardware resources. The suggested solution yields significant performance improvements over the state-of-the-art, achieving 59 frames per second (FPS) for GMM and 66 FPS for the Codebook method on full-HD (1080p) video streams. The results confirm the effectiveness of the proposed method in accelerating BS and demonstrate its suitability for real-time deployment in resource-constrained embedded environments. Full article

High-speed imaging offers a non-intrusive approach for monitoring falling-film flows over horizontal tube bundles, but reflective images are difficult to quantify because grayscale variations are jointly affected by film geometry, interfacial curvature, surface slope, viewing angle, and local highlights. This study proposes an interpretable visual-proxy sensing framework for comparative state assessment of such flows. Isothermal water experiments were conducted on a five-row horizontal tube bundle over Re_Γ = 184 − 960. For each condition, grayscale frames were acquired at fps and analyzed within five fixed row-wise regions of interest. The image sequence was transformed by temporal-median background subtraction, local spatiotemporal mapping, moving-average detrending, and median-absolute-deviation normalization. The resulting normalized map M_n and dynamic renewal field G were used to extract four scalar descriptors: noise-corrected apparent renewal intensity I_R, high-frequency fraction R_HF, spectral peak frequency f_p, and burst-event rate F_B. Results show that M_n and G capture the transition from sparse column flow to more continuous sheet flow and reveal row-dependent activity organization. The descriptors provide complementary information on renewal intensity, frequency composition, dominant time scale, and intermittent events. Zero-response, noise-correction, and sensitivity tests confirm that the framework avoids structured pseudo-waves and maintains stable row-wise comparisons. The method provides a low-calibration visual sensing tool for relative falling-film state assessment. Full article

Background: Optimal phase selection in coronary computed tomography angiography (CCTA) is crucial. While the mid-diastolic slow-filling (SF) phase is typically predicted using a conventional formula based on heart rate and atrioventricular conduction time, its validity in complete left bundle branch block (CLBBB)—where pronounced QRS prolongation induces severe mechanical dyssynchrony—remains unclear. We evaluated the impact of bundle branch block on cardiac-phase selection and validated a QRS-corrected predictive model. Methods: We retrospectively analyzed 94 patients (sinus rhythm, n = 40; complete right bundle branch block [CRBBB], n = 36; CLBBB, n = 18). Measured SF at the proximal right coronary artery was compared against predictions from the conventional formula (SF = −362 + 0.742 × [RR − PQ]) and a proposed QRS-corrected formula incorporating a “−(QRS − 100)” subtraction. To test the necessity of a novel model, regression analyses were reconstructed exclusively for the CLBBB cohort. Results: In CLBBB patients, the conventional formula critically overestimated SF by an average of 37.9 ms (RMSE 42.5 ms). Reconstructing simple and multivariate regression models exclusively for the CLBBB group yielded coefficients remarkably similar to the conventional formula, indicating that the fundamental physiological dependency on RR and PQ intervals remains intact despite the bundle branch block. Crucially, the simple proposed QRS-corrected formula successfully eliminated the overestimation bias (mean error −6.9 ms; p = 0.176) and demonstrated the highest predictive accuracy (RMSE 21.2 ms). Conclusions: A completely new predictive regression model is unnecessary for CLBBB patients. Simply incorporating a theoretical subtraction of pathological QRS prolongation optimally corrects the diastolic resting phase. Full article

On background cosmological scales, after subtraction of peculiar velocities and local bound-system motions, observed cosmological signals are redshifted rather than blueshifted. Yet, redshift alone does not distinguish the past lightcone of an expanding Universe from the future lightcone of a contracting one. In practice, the identification of the observed redshifted branch with the observational past is set primarily by electromagnetic radiation, whose retarded character is independently established in controlled physics, albeit over non-cosmological scales. From that perspective, the observed cosmological arrow is not separable from the causal/radiative prescription used to interpret the signals. This effective entanglement between the cosmological and the radiative arrows should nevertheless be distinguished from the notion of arrow used in the present work. Here instead, the relevant arrow is not thermodynamic but kinematic; it is defined by the symmetry or asymmetry of background lightcone observables under $\xi \leftrightarrow -\xi$, where $\xi \equiv ln(1+z)$ and z is the redshift—a criterion motivated directly by the time-reversal-symmetric special-relativistic longitudinal Doppler shift. Equivalently, the arrow considered here is the observed redshift/blueshift asymmetry of cosmological lightcone signals; retarded observations of an expanding FRW Universe are in the redshifted branch, whereas the opposite rapidity orientation would correspond to the blueshifted branch. This naturally suggests using rapidity-reversal symmetry as the redshift-space no-arrow condition when passing from special relativity (SR) to Friedmann–Robertson–Walker (FRW) cosmology, where the empty Milne Universe is a bridging borderline case. In fact, the viewpoint advocated here is that $\xi$-symmetry/asymmetry is practically more fundamental than t-symmetry/asymmetry simply because the former is more readily related to cosmological observables. It is shown here that generic non-empty FRW Universes possess an intrinsic $\xi$-asymmetry already at the background level, independently of entropy, coarse-graining, structure growth, or a Past Hypothesis. Full article

Background: Vertebral column resection (VCR) has historically been recognized as the most efficacious corrective intervention for severe rigid spinal deformities. Nevertheless, advancements in preoperative optimization, staged corrective methodologies, osteotomies, and contemporary instrumentation have broadened the spectrum of therapeutic options available. The definitive role of VCR in the modern management of rigid congenital spinal deformities remains a topic of ongoing scholarly discourse. Methods: This study presents two illustrative cases of severe congenital spinal deformities that were addressed employing various surgical methodologies, alongside a comprehensive review of the current literature pertaining to VCR and less invasive alternatives, including halo-gravity traction (HGT), temporary internal distraction techniques, pedicle subtraction osteotomy (PSO), asymmetric pedicle subtraction osteotomy (APSO), and multi-rod constructs. Results: The cases elucidated herein underscore the necessity for treatment strategies to be tailored specifically to the characteristics of the deformity, its flexibility, the neurological risks involved, and the individual patient’s specific attributes. In one case, significant deformity correction achieved via preoperative HGT facilitated successful management through multilevel Ponte osteotomies and posterior spinal fusion, thereby obviating the need for VCR. In other patient suffering from severe rigid congenital kyphotic deformity with pronounced anterior column deficiencies, VCR was deemed essential to realize adequate correction and neural decompression. All patients exhibited substantial radiographic correction, enhancements in health-related quality-of-life metrics, diminished disability and pain, while maintaining correction without neurological complications or implant failure at the final follow-up evaluation. Conclusions: VCR continues to be a vital element within the surgical repertoire for the treatment of severe rigid spinal deformities; however, it should not be deemed obligatory in every instance. Diligent preoperative evaluation, staged correction methodologies, and less invasive osteotomy techniques may permit satisfactory correction while mitigating surgical morbidity in suitably selected patients. Treatment approaches should be customized, favoring the least invasive procedure capable of achieving safe and lasting correction whenever practicable. Full article

Wearable-sensor human activity recognition (HAR) plays a key role in health monitoring, elderly care, and human–computer interaction. Deep learning dominates the field, but two limitations remain. CNNs with fixed kernels cannot capture cross-scale temporal events such as gait cycles and postural transitions in a single layer, and softmax attention on small sensor datasets is often diluted by common-mode background responses across the sequence. We propose DMSCNet, an end-to-end framework with two modules. The Dilated Multi-Scale Branch Block (DMSB) combines a shared bottleneck, parallel dilated convolutions, a pooling bypass, and SE-based channel recalibration to widen the temporal receptive field under a controlled parameter budget. The Contrastive Temporal Attention (CTA) module adopts a dual-path differential design, in which the two paths learn overlapping but non-identical attention patterns and their subtraction suppresses shared low-level responses while preserving the discriminative positions each path locks onto, encoded with opposite signs. DMSB and CTA are cascaded into a DMSC Block and stacked residually. On UCI-HAR, USC-HAD, and RealWorld, DMSCNet reaches F1-scores of 97.65%, 91.80%, and 99.05%, outperforming nine baselines. Ablations confirm that SE acts along the channel axis and CTA along the temporal axis, and visualization reveals a dynamic–static dichotomy together with a signed bipolar encoding pattern produced by the dual-path subtraction. Full article

The processing of two-dimensional (2D) spectral images constitutes a critical and multifaceted discipline in contemporary astronomical data analysis. As spectroscopic instruments evolve towards higher multiplexing, resolution, and sensitivity, the raw 2D data captured by detectors present increasingly complex challenges that transcend simple one-dimensional extraction. This review provides a systematic and comprehensive examination of the methodological evolution in this field over the past two decades. It gathered relevant studies by searching mainstream academic repositories and general search engines with the core keyword ‘2D Spectral Image’, and selected qualified references according to accessibility and research relevance. We categorize the landscape into three major paradigms: (1) physics-based modeling and algorithmic correction techniques for geometric distortion, scattered light, and sky background; (2) data-driven machine learning and deep learning approaches for image correction, spectral classification, and faint signal detection; and (3) the development of open-source software pipelines that democratize advanced processing. A central contribution of this review is a detailed comparative analysis of the performance metrics, underlying assumptions, and practical limitations of prominent algorithms. We highlight the transformative impact of convolutional neural networks (CNNs) and vision transformers (ViTs) on tasks such as celestial object classification and exoplanet detection, while also acknowledging the enduring importance of robust physical models for calibration and uncertainty quantification. The discussion culminates in an assessment of persistent challenges—including computational scalability, model generalizability, and interpretability—and outlines promising future directions at the intersection of AI, statistical inference, and large-scale survey science. Full article

Accurate detection of internal defects in rolls is crucial for industrial safety and product quality. Ultrasonic testing is a mainstream non-destructive method widely used for this purpose. However, in practice, ultrasonic echo signals often suffer from background clutter. When defects are located near the surface, weak defect echoes tend to couple with surface echoes, making signal extraction difficult and reducing the accuracy of subsequent feature extraction and classification. This paper proposes a novel ultrasonic signal preprocessing method aimed at improving the performance of subsequent defect identification models. The method first acquires ultrasonic signals from defect regions and background clutter reference signals from defect-free regions using a digital ultrasonic flaw detector. An improved median filter is then applied to remove spike interference and boundary outliers. On this basis, a multi-stage FIR (finite impulse response) filter is constructed, and particle swarm optimization is employed to adaptively optimize filter parameters, achieving an accurate estimation of background clutter. Finally, the clutter-suppressed defect signal is obtained through signal subtraction. Experimental results on a dataset of 5000 samples (2500 defective, 2500 non-defective) containing cylindrical artificial defects (diameter 8 mm, length 30 mm) demonstrate that using a CNN classifier with the same feature extraction and classification model, the signals preprocessed by the proposed method outperform traditional median filtering and wavelet denoising methods. The defect identification accuracy is improved by approximately 38 percentage points compared to median filtering and 20 percentage points compared to wavelet denoising, while also achieving a high recall rate, validating the effectiveness of the proposed method in enhancing roll internal defect detection. Full article

This paper proposes a cooled infrared focal plane circuit and detector imaging test system. The system features online acquisition, temperature monitoring, bias voltage programming, real-time imaging, and online testing capabilities, improving the testing and verification efficiency of infrared focal plane detector imaging systems. Temperature monitoring accuracy reaches ±0.02 K; bias voltage programming keeps bias noise below 100 nV/Hz^1/2; and a 2 × 2 pixel image array is used for data transmission, enabling algorithmic computation on pixel data for convenient image processing. Furthermore, this paper proposes various image algorithms and overall algorithm structures based on a single-chip pixel-level ADC, the dynamic range is 141.8 dB, and the maximum input charge capacity is 4.46 Ge⁻. The overall structure of the multifunctional image algorithm for infrared focal plane detectors is summarized, including background subtraction, blind pixel compensation, non-uniformity correction, windowing, pixel merging, spatial filtering, histogram equalization, and time delay integration (TDI). Non-uniformity is improved by 98.3%, blind pixel rate is reduced by 92.3%, background subtraction performance is improved by 93%, and spatial filtering and TDI improve relative spatial noise by 63% and 89%, respectively. Pixel merging can increase the gray mean by 4.09 times, enabling arbitrary windowing and uniform histogram distribution. These highly practical research efforts will drive the further development of future larger-array cooled infrared focal plane array (IRFPA) detector imaging test technology and more intelligent infrared image algorithms. Full article

Background/Objectives: Multiplex and point-of-care (POC) diagnostics require each probe to detect one intended target while rejecting many closely related sequences under shared room-temperature conditions. The conventional focus on mismatch count is incomplete: two alignments with the same number of matches and mismatches can have very different off-target risks depending on whether mismatches are clustered or distributed. We introduce a simple visual heuristic that scores mismatch placement rather than mismatch count alone. Methods: Effective complementary island (ECI) score retained matched continuity after subtracting one base for each mismatch- or gap-exposed edge. The score is S_ECI = Σ_i ECI_i^2, and the design margin is ΔS_ECI = S_ECI (intended) − S_ECI (highest-scoring non-intended alignment by ECI). ECI is not a thermodynamic model; thermodynamics (ΔG37) is used separately to verify an adequate sensitivity floor. We retrospectively applied ECI to a fixed 39-target HPV capture-probe benchmark and to a public Affymetrix dataset contrasting clustered versus distributed mismatches at identical or near-identical mismatch counts. Results: In the HPV benchmark, ECI separated intended from off-target in 32/39 panels; ΔG37 favored the intended duplex in 31/39 panels; both layers were concordant in 36/39 panels. In the Affymetrix dataset (n = 8 probes, 2–4 mismatches), S_ECI correlated with reported log2 hybridization intensity (Pearson r = 0.92, p = 0.0014). Within the strict three-mismatch subset (n = 5), S_ECI remained correlated with intensity (r = 0.96; p = 0.010), while ΔG37 was uncorrelated (r = −0.04; p = 0.95), supporting the narrower claim that mismatch placement can affect signal even when mismatch count is fixed. Conclusions: ECI is not a replacement for thermodynamics, BLAST, target-accessibility analysis, empirical optimization, or machine-learning prediction. It adds one actionable readout: where to shift, shorten, or place a limited intentional mismatch so that intended retained continuity stays above the assay floor while the highest-scoring off-target island by ECI is fragmented. We provide a bench-ready workflow for multiplex, room-temperature, and POC probe design. Full article

Background/Objectives: Endovascular treatment of cerebral arteriovenous malformations (AVMs) and arteriovenous fistulas (AVFs) is associated with substantial radiation exposure due to procedural complexity and repeated angiographic acquisitions. This study evaluates radiation exposure during AVM and AVF embolization and establishes local diagnostic reference levels (DRLs). Methods: A single-center retrospective dose audit was conducted, encompassing 114 endovascular procedures performed using a low-dose workflow. Radiation exposure was quantified using dose area product (DAP), reference air kerma (K_a,r), fluoroscopy time (FT), and the number of digital subtraction angiography (DSA) frames per procedure. Median values were defined as the median (P50), and local DRLs as the 75th percentile (P75). Comparative analyses were conducted between AVM and AVF procedures, between male and female patients, and within selected AVM subgroups. Results: The analysis comprised 86 AVM procedures and 28 AVF procedures. For AVMs, the local DRLs (P75) were 28.9 Gy·cm² for DAP, 400 mGy for K_a,r, 310 DSA frames per procedure, and 1619 s for FT. For AVFs, the respective values were 47.3 Gy·cm², 465 mGy, 478 DSA frames, and 1820 s. No statistically significant differences were identified between female and male patients. However, AVF procedures demonstrated significantly higher radiation exposure than AVM procedures for all parameters except FT. Within the AVM subgroup, no significant differences were observed between single-stage and other AVM procedures or between female and male patients. Conclusions: AVM and AVF embolization procedures are dose-intensive neuroendovascular interventions. Establishing local DRLs for AVM and AVF may enhance radiation monitoring and facilitate procedure-specific dose optimization. Full article

Background and Clinical Significance: Bilateral carotid-cavernous fistulas are rare clinical entities characterized by heterogeneous clinical presentations and variable outcomes. Case presentation: We report the case of a 69-year-old woman with a three-month history of progressive bilateral conjunctival hyperemia, proptosis, intermittent diplopia, and a left eye abduction deficit. Her systemic history included long-standing arterial hypertension and previous thyroidectomy with stable substitutive therapy. Comprehensive ophthalmologic, neurologic, and endocrine evaluations excluded more common causes of orbital congestion, including thyroid eye disease, orbital cellulitis, cavernous sinus thrombosis, and idiopathic orbital inflammation. The patient denied any history of recent trauma. Digital subtraction angiography (DSA) confirmed a bilateral, low-flow, indirect Barrow type D carotid–cavernous fistula (CCF) supplied by dural branches of both the internal and external carotid arteries, with marked reflux into dilated superior ophthalmic veins. DSA was essential, as prior CT and MRI studies did not identify any vascular abnormalities. The patient was scheduled for transvenous embolization; however, during the follow-up she noted gradual improvement in her condition. Repeat pre-procedural angiography performed approximately two months later demonstrated complete spontaneous closure of all shunts, accompanied by full clinical resolution. Conclusions: Owing to the exceptional rarity of bilateral indirect CCFs and the added occurrence of spontaneous closure, this case expands the limited existing literature and emphasizes the diagnostic challenges and the need for individualized treatment timing supported by multidisciplinary evaluation in low-flow dural carotid–cavernous fistulas. Full article

Background/Objectives: How the histologically benign tier of lung preinvasive lesions—atypical adenomatous hyperplasia (AAH) and squamous dysplasia (SD)—relates genomically to its paired carcinoma is unclear. To identify early versus late events, we compared synchronous preinvasive and invasive lesions from the same patient. Methods: Whole-exome sequencing was performed on 33 FFPE samples from 11 patients (7 AAH–lung adenocarcinoma [LUAD] and 4 SD–squamous cell carcinoma [SqCC] pairs, with paired normal lung). FFPE artefacts were mitigated by paired-normal subtraction, panel-of-normals filtering, and orthogonal caller cross-validation. Cancer-panel variants were classified as cancer-only, shared, or preinvasive-only. Results: Only ∼10% of cancer-panel variants were shared between paired lesions (∼50% carcinoma-only, ∼40% preinvasive-only), indicating that benign AAH/SD do not broadly mirror the paired carcinoma. Within this small shared fraction, the early-driver pattern diverged between tracks: AAH–LUAD pairs tended to share EGFR alterations, whereas the four SD–SqCC pairs featured MET-pathway alterations without any shared EGFR events. TP53 and most other canonical drivers were carcinoma-confined (within-cohort contrast direction-consistent but non-significant); three patients lacked any canonical driver despite substantial mutational burden. Conclusions: In this pilot cohort, benign AAH and SD were genomically largely distinct from their paired carcinomas, sharing only a small set of key drivers whose identity diverged between glandular and squamous tracks. This suggests that benign-appearing AAH/SD differ from the more advanced AIS/MIA precursors not only histologically but also at the genomic level. These hypothesis-generating findings require confirmation in larger, multi-omic cohorts. Full article

Background/Objectives: Differentiating focal nodular hyperplasia (FNH) from hepatic adenoma (HA) remains challenging, as FNH is benign whereas HA carries risks of hemorrhage and malignant transformation. This prospective single-center pilot study evaluated the diagnostic performance of three-dimensional magnetic resonance elastography (3D-MRE) using a gravitational transducer for non-invasive differentiation of FNH and HA. Methods: Thirty-three participants (23 FNH, 10 HA) underwent 3D-MRE using the gravitational transducer. Viscoelastic parameters—stiffness, shear wave speed (Cs), wave attenuation, and phase angle—were quantified for lesions and background parenchyma. $\Delta$-values were calculated by subtracting background liver measurements from lesion values. Results: FNH demonstrated significantly higher stiffness than HA (median 3.16 vs. 2.58 kPa; p = 0.02) and higher Cs (median 1.81 vs. 1.64 m/s; p = 0.001). Normalized stiffness differences ($\Delta$ stiffness) were significantly greater in FNH than HA (median 0.83 vs. 0.10 kPa; p = 0.001). Generalized additive models revealed divergent volume-dependent stiffening behaviors. In ROC analysis, $\Delta$ stiffness and $\Delta$ Cs each achieved an AUC of 0.87, indicating that single background-normalized viscoelastic parameters carry the principal diagnostic signal. An exploratory multivariable combination of $\Delta$ stiffness with patient age produced an apparent AUC of 0.93 with wide odds-ratio confidence intervals, and is presented as hypothesis-generating rather than as a clinical prediction model. Conclusions: In this pilot cohort, 3D-MRE using the gravitational transducer showed encouraging parameter-level separation between FNH and HA, with background normalization enhancing discrimination. Wave attenuation and phase angle did not differ significantly between lesion types. Given the small sample size (particularly the HA subgroup of ten patients), the mixed reference standard (histological confirmation in only 14 of 33 lesions; definitive hepatobiliary-phase MRI criteria in 19 of 33), the single-slice ROI used for lesion measurement, and the incomplete characterization of background liver parenchyma, these findings should be regarded as hypothesis-generating and require external validation in larger, multicenter cohorts before any clinical application. Full article

Background and Objectives: Pericoronary adipose tissue (PCAT) attenuation measured on coronary CT angiography is a promising imaging biomarker of coronary inflammation; however, absolute values may be influenced by technical and inter-individual variability, and a standardized methodology for measurement has not been established. Our study aimed to evaluate the association between PCAT attenuation and CAC burden while comparing absolute attenuation values with normalized values to minimize these sources of variability. Materials and Methods: Two hundred patients undergoing cardiac CT were included and stratified into four CAC categories (0, 1–99, 100–299, ≥300). PCAT attenuation was measured at multiple locations on two main levels: aortic root level and four-chamber view level. Relative PCAT attenuation was calculated by subtracting subcutaneous fat attenuation from raw PCAT values. Group comparisons were performed using ANOVA or Kruskal–Wallis tests, and multivariable linear regression models were adjusted for age, sex, and body mass index. Results: In univariate analysis, relative PCAT attenuation differed significantly across CAC categories at the aortic-level right coronary artery (RCA) site (p = 0.007). In multivariable analysis, higher CAC categories were associated with increased relative PCAT attenuation at the aortic RCA (β = 8.56, p = 0.015 for CAC 100–299; β = 10.68, p = 0.005 for CAC ≥300), while associations at the left main coronary artery (LMCA) showed significance in low and moderate CAC categories (β = 6.91, p = 0.047 for CAC 1–99 and β = 8.57, p = 0.016 for CAC 100–299). No significant associations were observed between CAC and raw PCAT attenuation at the aortic level, while isolated and inconsistent findings were observed in other territories. Conclusions: Relative PCAT attenuation is independently associated with CAC severity and normalized values may reduce technical and biological variability, potentially enhancing the sensitivity and robustness of this CT-based biomarker. Full article