RESUMEN
(1) Background: The aim of this study was to evaluate short- to mid-term clinical and radiological results in patients undergoing primary total hip arthroplasty (THA) with the use of a Selective Laser Melting 3D-printed highly porous titanium acetabular cup (Jump System Traser®, Permedica Orthopaedics). (2) Methods: We conducted a retrospective study and collected prospective data on 125 consecutive patients who underwent primary THA with the use of highly porous titanium cup. Each patient was evaluated preoperatively and postoperatively with a clinical and radiological assessment. (3) Results: The final cohort consisted of 104 patients evaluated after a correct value of 52 (38-74) months. The median Harris Hip Score (HHS) significantly improved from 63.7 (16-95.8) preoperatively to 94.8 (38.2-95.8) postoperatively (p < 0.001), with higher improvement associated with higher age at surgery (ß = 0.22, p = 0.025). On postoperative radiographs, the average acetabular cup inclination and anteversion were 46° (30°-57°) and 15° (1°-32°), respectively. All cups radiographically showed signs of osseointegration with no radiolucency observed, or component loosening. (4) Conclusions: The use of this highly porous acetabular cup in primary THA achieved excellent clinical, functional, and radiological results at mid-term follow-up. A better clinical recovery can be expected in older patients. The radiological evaluation showed excellent osseointegration of the cup with complete absence of periprosthetic radiolucent lines.
RESUMEN
Background: The medium-term results of reverse shoulder arthroplasty (RSA) that has been performed by a single surgeon have been previously reported. The purpose of this study was to investigate the minimum 10-year clinical and radiographic outcomes of these patients. Methods: In this prospective cohort study, 27 patients were evaluated after RSA for massive rotator cuff tear with or without eccentric osteoarthritis (OA) or concentric OA with the Constant-Murley Score (CMS), range of motion (ROM), and a radiologic assessment. Results: At a mean 12-year follow-up, the CMS and ROM were significantly improved when compared with the baseline values (all P < .001). Once stratified by diagnosis, no difference in the ROM or total CMS was found between patients with massive rotator cuff tear with/without eccentric OA and those with concentric OA. Neither ROM nor CMS decreased when compared to the mid-term values of the previous study, for both the overall population and the diagnosis-stratified groups. Scapular notching was reported in 66.7% of cases that was similar to the data reported at mid-term follow-up. The calcification rate was 59.3% at the long-term evaluation, and there were no differences between the same case-series population (51.9%; P = .785) and the whole population at mid-term follow-up (47%; P = .358). Conclusion: RSA led to excellent clinical and functional outcomes for patients up to 17 years postoperatively, and there was no decrease in the CMS over time. No loosening of implants was noted, and the rate of scapular notching was 66%, mostly grade 1 or 2.
RESUMEN
BACKGROUND: Several techniques have been adopted during primary reverse shoulder arthroplasty (RSA) to manage glenoid bone defect. Among bone grafts, humeral head autograft is currently the mainstream option. However, autologous humeral heads may be unavailable or inadequate, and allografts may be a viable alternative. The aim of the present study was to evaluate the functional and radiologic outcomes of femoral head allografts for glenoid bone defects in primary RSA. METHODS: We conducted a retrospective study with prospective data collection enrolling 20 consecutive patients who underwent RSA with femoral head allografts for glenoid bone defects. Indications for surgery were eccentric cuff tear arthropathy in 10 cases (50%), concentric osteoarthritis in 9 cases (45%), and fracture sequelae in 1 case (5%). Each patient was evaluated preoperatively and at follow-up by radiologic and computed tomography (CT) and by assessing the range of motion (ROM) and the Constant-Murley score (CMS). A CT-based software, a patient-specific 3D model of the scapula, and patient-specific instrumentation were used to shape the graft and to assess the position of K-wire for the central peg. Postoperatively, CT scans were used to identify graft incorporation and resorption. RESULTS: After a median follow-up of 26.5 months (24-38), ROM and CMS showed a statistically significant improvement (all P = .001). The median measures of the graft were as follows: 28 mm (28-29) for diameter, 22° (10°-31°) for angle, 4 mm (2-8 mm) for minimum thickness, and 15 mm (11-21 mm) for maximum thickness. Before the surgery, the median glenoid version was 21.8° (16.5°-33.5°) for the retroverted glenoids and -13.5° (-23° to -12°) for the anteverted glenoids. At the follow-up, the median postoperative baseplate retroversion was 5.7° (2.2°-1.5°) (P = .001), and this value was close to the 4° retroversion planned on the preoperative CT-based software. Postoperative major complications were noted in 4 patients: 2 dislocations, 1 baseplate failure following a high-energy trauma, and 1 septic baseplate failure. Partial graft resorption without glenoid component failure was observed in 3 cases that did not require revision surgery. CONCLUSION: The femoral head allograft for glenoid bone loss in primary RSA restores shoulder function, with CMS values comparable to those of sex- and age-matched healthy individuals. A high rate of incorporation of the graft and satisfactory correction of the glenoid version can be expected after surgery. The management of glenoid bone defects remains a challenging procedure, and a 15% risk of major complication must be considered.
Asunto(s)
Artroplastía de Reemplazo de Hombro , Cavidad Glenoidea , Articulación del Hombro , Humanos , Articulación del Hombro/diagnóstico por imagen , Articulación del Hombro/cirugía , Artroplastía de Reemplazo de Hombro/efectos adversos , Cabeza Femoral/trasplante , Estudios Retrospectivos , Escápula/cirugía , Complicaciones Posoperatorias/etiología , Aloinjertos/cirugía , Cavidad Glenoidea/diagnóstico por imagen , Cavidad Glenoidea/cirugía , Resultado del Tratamiento , Rango del Movimiento ArticularRESUMEN
BACKGROUND: Literature describing outcomes and integrity after isolated subscapularis (SSC) tendon repair is emerging but remains limited to a few small case series with short-term follow-up. The aim of this study was to evaluate the long-term clinical outcomes and repair integrity in patients who underwent arthroscopic repair of isolated SSC tears. METHODS: A retrospective study was conducted with the following inclusion criteria: (1) primary and elective shoulder arthroscopy for isolated SSC repair, (2) type III (a full-thickness tear in the upper two-thirds of the tendon) or IV (a complete tear without tendon retraction) SSC tear according to the Lafosse classification, and (3) a minimum 24-month follow-up. Preoperatively, the range of motion (ROM) and the Constant-Murley score (CMS) and at follow-up, the ROM, the University of California-Los Angeles (UCLA) Shoulder Rating Scale, the Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire, and the CMS were evaluated; an ultrasonographic assessment of tendon healing was performed according to the Sugaya classification. RESULTS: The final sample consisted of 45 patients with an average age of 55 ± 9 years. After a mean follow-up time of 107 ± 54 months, the mean UCLA and DASH scores were 8.7 ± 1.3 and 42.2 ± 6.4, respectively. ROM and CMS showed statistically significant improvements (all P < .001). Before surgery, the mean CMS was 49% that of sex- and age-matched healthy individuals, and all patients showed a CMS lower than the normative data. At the final follow-up visit, the mean CMS was 94.2% that of sex- and age-matched healthy individuals, and no patients showed CMS of 30 or less. The mean increase in the CMS was 41.4 ± 9.8 points (range, 23-60 points). The ultrasonographic assessment showed SSC tendon healing in 39 (86.7%) cases; tendon retear was recorded in 5 (13.3%) cases. All scores directly correlated with the healing of the tendon. A higher postoperative DASH score was associated with male sex (P = .039, ß = 5.538) and a longer follow-up period (P = .044, ß = 0.001). The postoperative CMS (P < .001) and UCLA scores (P = .001) were significantly higher in patients younger than 60 years of age at surgery than in older individuals. CONCLUSION: Arthroscopic repair of isolated SSC tears achieves excellent clinical and functional results at a mean of 9 years postoperatively, with a satisfactory healing rate. Better functional outcomes correlate with SSC tendon integrity and were observed in male patients and in those younger than 60 years at surgery.
RESUMEN
Many pathogens exploit host cell-surface glycans. However, precise analyses of glycan ligands binding with heavily modified pathogen proteins can be confounded by overlapping sugar signals and/or compounded with known experimental constraints. Universal saturation transfer analysis (uSTA) builds on existing nuclear magnetic resonance spectroscopy to provide an automated workflow for quantitating protein-ligand interactions. uSTA reveals that early-pandemic, B-origin-lineage severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spike trimer binds sialoside sugars in an "end-on" manner. uSTA-guided modeling and a high-resolution cryo-electron microscopy structure implicate the spike N-terminal domain (NTD) and confirm end-on binding. This finding rationalizes the effect of NTD mutations that abolish sugar binding in SARS-CoV-2 variants of concern. Together with genetic variance analyses in early pandemic patient cohorts, this binding implicates a sialylated polylactosamine motif found on tetraantennary N-linked glycoproteins deep in the human lung as potentially relevant to virulence and/or zoonosis.
Asunto(s)
COVID-19 , Interacciones Huésped-Patógeno , SARS-CoV-2 , Ácidos Siálicos , Glicoproteína de la Espiga del Coronavirus , COVID-19/transmisión , Microscopía por Crioelectrón , Variación Genética , Humanos , Resonancia Magnética Nuclear Biomolecular , Polisacáridos/química , Unión Proteica , Dominios Proteicos , SARS-CoV-2/química , SARS-CoV-2/genética , Ácidos Siálicos/química , Glicoproteína de la Espiga del Coronavirus/química , Glicoproteína de la Espiga del Coronavirus/genéticaRESUMEN
Graph drawing techniques have been developed in the last few years with the purpose of producing esthetically pleasing node-link layouts. Recently, the employment of differentiable loss functions has paved the road to the massive usage of gradient descent and related optimization algorithms. In this article, we propose a novel framework for the development of Graph Neural Drawers (GNDs), machines that rely on neural computation for constructing efficient and complex maps. GND is Graph Neural Networks (GNNs) whose learning process can be driven by any provided loss function, such as the ones commonly employed in Graph Drawing. Moreover, we prove that this mechanism can be guided by loss functions computed by means of feedforward neural networks, on the basis of supervision hints that express beauty properties, like the minimization of crossing edges. In this context, we show that GNNs can nicely be enriched by positional features to deal also with unlabeled vertexes. We provide a proof-of-concept by constructing a loss function for the edge crossing and provide quantitative and qualitative comparisons among different GNN models working under the proposed framework.
RESUMEN
The combined impact of common and rare exonic variants in COVID-19 host genetics is currently insufficiently understood. Here, common and rare variants from whole-exome sequencing data of about 4000 SARS-CoV-2-positive individuals were used to define an interpretable machine-learning model for predicting COVID-19 severity. First, variants were converted into separate sets of Boolean features, depending on the absence or the presence of variants in each gene. An ensemble of LASSO logistic regression models was used to identify the most informative Boolean features with respect to the genetic bases of severity. The Boolean features selected by these logistic models were combined into an Integrated PolyGenic Score that offers a synthetic and interpretable index for describing the contribution of host genetics in COVID-19 severity, as demonstrated through testing in several independent cohorts. Selected features belong to ultra-rare, rare, low-frequency, and common variants, including those in linkage disequilibrium with known GWAS loci. Noteworthily, around one quarter of the selected genes are sex-specific. Pathway analysis of the selected genes associated with COVID-19 severity reflected the multi-organ nature of the disease. The proposed model might provide useful information for developing diagnostics and therapeutics, while also being able to guide bedside disease management.
Asunto(s)
COVID-19/genética , COVID-19/fisiopatología , Secuenciación del Exoma , Predisposición Genética a la Enfermedad , Fenotipo , Índice de Severidad de la Enfermedad , Adulto , Anciano , Anciano de 80 o más Años , Estudios de Cohortes , Femenino , Alemania , Humanos , Italia , Masculino , Persona de Mediana Edad , Polimorfismo de Nucleótido Simple , Quebec , SARS-CoV-2 , Suecia , Reino UnidoRESUMEN
Adversarial attacks on machine learning-based classifiers, along with defense mechanisms, have been widely studied in the context of single-label classification problems. In this paper, we shift the attention to multi-label classification, where the availability of domain knowledge on the relationships among the considered classes may offer a natural way to spot incoherent predictions, i.e., predictions associated to adversarial examples lying outside of the training data distribution. We explore this intuition in a framework in which first-order logic knowledge is converted into constraints and injected into a semi-supervised learning problem. Within this setting, the constrained classifier learns to fulfill the domain knowledge over the marginal distribution, and can naturally reject samples with incoherent predictions. Even though our method does not exploit any knowledge of attacks during training, our experimental analysis surprisingly unveils that domain-knowledge constraints can help detect adversarial examples effectively, especially if such constraints are not known to the attacker. We show how to implement an adaptive attack exploiting knowledge of the constraints and, in a specifically-designed setting, we provide experimental comparisons with popular state-of-the-art attacks. We believe that our approach may provide a significant step towards designing more robust multi-label classifiers.
RESUMEN
Symmetries, invariances and conservation equations have always been an invaluable guide in Science to model natural phenomena through simple yet effective relations. For instance, in computer vision, translation equivariance is typically a built-in property of neural architectures that are used to solve visual tasks; networks with computational layers implementing such a property are known as Convolutional Neural Networks (CNNs). This kind of mathematical symmetry, as well as many others that have been recently studied, are typically generated by some underlying group of transformations (translations in the case of CNNs, rotations, etc.) and are particularly suitable to process highly structured data such as molecules or chemical compounds which are known to possess those specific symmetries. When dealing with video streams, common built-in equivariances are able to handle only a small fraction of the broad spectrum of transformations encoded in the visual stimulus and, therefore, the corresponding neural architectures have to resort to a huge amount of supervision in order to achieve good generalization capabilities. In the paper we formulate a theory on the development of visual features that is based on the idea that movement itself provides trajectories on which to impose consistency. We introduce the principle of Material Point Invariance which states that each visual feature is invariant with respect to the associated optical flow, so that features and corresponding velocities are an indissoluble pair. Then, we discuss the interaction of features and velocities and show that certain motion invariance traits could be regarded as a generalization of the classical concept of affordance. These analyses of feature-velocity interactions and their invariance properties leads to a visual field theory which expresses the dynamical constraints of motion coherence and might lead to discover the joint evolution of the visual features along with the associated optical flows.
RESUMEN
Sketching is a universal communication tool that, despite its simplicity, is able to efficiently express a large variety of concepts and, in some limited contexts, it can be even more immediate and effective than natural language. In this paper we explore the feasibility of using neural networks to approach sketching in the same way they are commonly used in Language Modeling. We propose a novel approach to what we refer to as "Sketch Modeling", in which a neural network is exploited to learn a probabilistic model that estimates the probability of sketches. We focus on simple sketches and, in particular, on the case in which sketches are represented as sequences of segments. Segments and sequences can be either given - when the sketches are originally drawn in this format - or automatically generated from the input drawing by means of a procedure that we designed to create short sequences, loosely inspired by the human behavior. A Recurrent Neural Network is used to learn the sketch model and, afterward, the network is seeded with an incomplete sketch that it is asked to complete, generating one segment at each time step. We propose a set of measures to evaluate the outcome of a Beam Search-based generation procedure, showing how they can be used to identify the most promising generations. Our experimental analysis assesses the feasibility of this way of modeling sketches, also in the case in which several different categories of sketches are considered.
Asunto(s)
Lenguaje , Redes Neurales de la Computación , Humanos , ProbabilidadRESUMEN
BACKGROUND: While SARS-CoV-2 similarly infects men and women, COVID-19 outcome is less favorable in men. Variability in COVID-19 severity may be explained by differences in the host genome. METHODS: We compared poly-amino acids variability from WES data in severely affected COVID-19 patients versus SARS-CoV-2 PCR-positive oligo-asymptomatic subjects. FINDINGS: Shorter polyQ alleles (≤22) in the androgen receptor (AR) conferred protection against severe outcome in COVID-19 in the first tested cohort (both males and females) of 638 Italian subjects. The association between long polyQ alleles (≥23) and severe clinical outcome (p = 0.024) was also validated in an independent cohort of Spanish men <60 years of age (p = 0.014). Testosterone was higher in subjects with AR long-polyQ, possibly indicating receptor resistance (p = 0.042 Mann-Whitney U test). Inappropriately low serum testosterone level among carriers of the long-polyQ alleles (p = 0.0004 Mann-Whitney U test) predicted the need for intensive care in COVID-19 infected men. In agreement with the known anti-inflammatory action of testosterone, patients with long-polyQ and age ≥60 years had increased levels of CRP (p = 0.018, not accounting for multiple testing). INTERPRETATION: We identify the first genetic polymorphism that appears to predispose some men to develop more severe disease. Failure of the endocrine feedback to overcome AR signaling defects by increasing testosterone levels during the infection leads to the polyQ tract becoming dominant to serum testosterone levels for the clinical outcome. These results may contribute to designing reliable clinical and public health measures and provide a rationale to test testosterone as adjuvant therapy in men with COVID-19 expressing long AR polyQ repeats. FUNDING: MIUR project "Dipartimenti di Eccellenza 2018-2020" to Department of Medical Biotechnologies University of Siena, Italy (Italian D.L. n.18 March 17, 2020) and "Bando Ricerca COVID-19 Toscana" project to Azienda Ospedaliero-Universitaria Senese. Private donors for COVID-19 research and charity funds from Intesa San Paolo.
Asunto(s)
COVID-19/patología , Péptidos/genética , Receptores Androgénicos/genética , Anciano , Estudios de Casos y Controles , Cuidados Críticos/estadística & datos numéricos , Femenino , Genoma Humano/genética , Humanos , Masculino , Persona de Mediana Edad , Polimorfismo de Nucleótido Simple/genética , Factores de Riesgo , SARS-CoV-2 , Índice de Severidad de la Enfermedad , España , Testosterona/sangreRESUMEN
Within the GEN-COVID Multicenter Study, biospecimens from more than 1000 SARS-CoV-2 positive individuals have thus far been collected in the GEN-COVID Biobank (GCB). Sample types include whole blood, plasma, serum, leukocytes, and DNA. The GCB links samples to detailed clinical data available in the GEN-COVID Patient Registry (GCPR). It includes hospitalized patients (74.25%), broken down into intubated, treated by CPAP-biPAP, treated with O2 supplementation, and without respiratory support (9.5%, 18.4%, 31.55% and 14.8, respectively); and non-hospitalized subjects (25.75%), either pauci- or asymptomatic. More than 150 clinical patient-level data fields have been collected and binarized for further statistics according to the organs/systems primarily affected by COVID-19: heart, liver, pancreas, kidney, chemosensors, innate or adaptive immunity, and clotting system. Hierarchical clustering analysis identified five main clinical categories: (1) severe multisystemic failure with either thromboembolic or pancreatic variant; (2) cytokine storm type, either severe with liver involvement or moderate; (3) moderate heart type, either with or without liver damage; (4) moderate multisystemic involvement, either with or without liver damage; (5) mild, either with or without hyposmia. GCB and GCPR are further linked to the GCGDR, which includes data from whole-exome sequencing and high-density SNP genotyping. The data are available for sharing through the Network for Italian Genomes, found within the COVID-19 dedicated section. The study objective is to systematize this comprehensive data collection and begin identifying multi-organ involvement in COVID-19, defining genetic parameters for infection susceptibility within the population, and mapping genetically COVID-19 severity and clinical complexity among patients.
Asunto(s)
Bancos de Muestras Biológicas , COVID-19/genética , Predisposición Genética a la Enfermedad , Sistema de Registros , SARS-CoV-2 , Manejo de Especímenes , Adolescente , Adulto , COVID-19/epidemiología , Femenino , Humanos , Italia , MasculinoRESUMEN
BACKGROUND: Timely recognition of malignant melanoma (MM) is challenging for dermatologists worldwide and represents the main determinant for mortality. Dermoscopic examination is influenced by dermatologists' experience and fails to achieve adequate accuracy and reproducibility in discriminating atypical nevi (AN) from early melanomas (EM). OBJECTIVE: We aimed to develop a Deep Convolutional Neural Network (DCNN) model able to support dermatologists in the classification and management of atypical melanocytic skin lesions (aMSL). METHODS: A training set (630 images), a validation set (135) and a testing set (214) were derived from the idScore dataset of 979 challenging aMSL cases in which the dermoscopic image is integrated with clinical data (age, sex, body site and diameter) and associated with histological data. A DCNN_aMSL architecture was designed and then trained on both dermoscopic images of aMSL and the clinical/anamnestic data, resulting in the integrated "iDCNN_aMSL" model. Responses of 111 dermatologists with different experience levels on both aMSL classification (intuitive diagnosis) and management decisions (no/long follow-up; short follow-up; excision/preventive excision) were compared with the DCNNs models. RESULTS: In the lesion classification study, the iDCNN_aMSL achieved the best accuracy, reaching an AUC = 90.3 %, SE = 86.5 % and SP = 73.6 %, compared to DCNN_aMSL (SE = 89.2 %, SP = 65.7 %) and intuitive diagnosis of dermatologists (SE = 77.0 %; SP = 61.4 %). CONCLUSIONS: The iDCNN_aMSL proved to be the best support tool for management decisions reducing the ratio of inappropriate excision. The proposed iDCNN_aMSL model can represent a valid support for dermatologists in discriminating AN from EM with high accuracy and for medical decision making by reducing their rates of inappropriate excisions.
Asunto(s)
Aprendizaje Profundo , Dermoscopía/métodos , Interpretación de Imagen Asistida por Computador/métodos , Melanoma/diagnóstico , Nevo/diagnóstico , Neoplasias Cutáneas/diagnóstico , Adolescente , Adulto , Anciano , Anciano de 80 o más Años , Niño , Preescolar , Conjuntos de Datos como Asunto , Diagnóstico Diferencial , Femenino , Humanos , Lactante , Recién Nacido , Masculino , Persona de Mediana Edad , Reproducibilidad de los Resultados , Estudios Retrospectivos , Piel/diagnóstico por imagen , Adulto JovenRESUMEN
PURPOSE: To investigate associations between clinical and demographic parameters and Constant-Murley (CM) scores after subacromial balloon placement for massive, irreparable rotator cuff tears and to evaluate implant survival, shoulder function, and patient satisfaction. METHODS: We prospectively analyzed patients with rotator cuff tears deemed irreparable on preoperative magnetic resonance imaging for whom nonoperative therapy was unsuccessful and who underwent balloon placement from 2014 to 2017 with minimum 1-year follow-up. Shoulder function was assessed using the CM score and the 12-Item Short Form Health Survey. RESULTS: The study included 51 patients (22 women and 29 men) with a mean age at surgery of 63 years (range, 50-78 years). The mean follow-up period was 36 months (range, 24-56 months). The postoperative acromiohumeral interval and total preoperative CM score predicted the postoperative CM score at final follow-up. The implant survival rates were 92% at 6 and 12 months, 90% at 2 years, and 87% at 3 and 4 years. Five patients underwent reverse total shoulder arthroplasty, and 1 underwent latissimus dorsi tendon transfer. Postoperatively, mean CM scores (± standard deviation) improved for range of motion (from 11 ± 5.4 to 34 ± 6.8) and strength (from 13 ± 5.4 to 28 ± 12) (P < .01 for both). The total CM score improved from 27 ± 7.4 preoperatively to 77 ± 15 postoperatively (P < .01). The physical and mental component summary scores on the 12-Item Short Form Health Survey improved from 27 ± 5.0 to 51 ± 6.5 (P = .02) and from 44 ± 15 to 56 ± 8.0, respectively (P < .01). Thirty-eight patients reported excellent satisfaction, 8 were satisfied, and 5 were dissatisfied. Of the patients, 50 (98%) exceeded the minimal clinically important difference (≥10.4) and patient acceptable symptom state (≥44). CONCLUSIONS: At mean 3-year follow-up, subacromial balloon spacer placement for massive, irreparable rotator cuff tears was associated with a significant improvement in shoulder function, limited need for revision surgery, and high patient satisfaction. A greater postoperative acromiohumeral interval and lower preoperative CM score predicted a lower postoperative CM score at final follow-up. LEVEL OF EVIDENCE: Level IV, retrospective cohort study and treatment study.
Asunto(s)
Satisfacción del Paciente , Lesiones del Manguito de los Rotadores/cirugía , Hombro/fisiopatología , Hombro/cirugía , Anciano , Anciano de 80 o más Años , Femenino , Estudios de Seguimiento , Humanos , Imagen por Resonancia Magnética , Masculino , Persona de Mediana Edad , Periodo Posoperatorio , Cuidados Preoperatorios , Rango del Movimiento Articular , Reoperación , Estudios Retrospectivos , Manguito de los Rotadores/patología , Manguito de los Rotadores/cirugía , Lesiones del Manguito de los Rotadores/diagnóstico por imagen , Articulación del Hombro/cirugía , Resultado del TratamientoRESUMEN
Visual attention refers to the human brain's ability to select relevant sensory information for preferential processing, improving performance in visual and cognitive tasks. It proceeds in two phases. One in which visual feature maps are acquired and processed in parallel. Another where the information from these maps is merged in order to select a single location to be attended for further and more complex computations and reasoning. Its computational description is challenging, especially if the temporal dynamics of the process are taken into account. Numerous methods to estimate saliency have been proposed in the last 3 decades. They achieve almost perfect performance in estimating saliency at the pixel level, but the way they generate shifts in visual attention fully depends on winner-take-all (WTA) circuitry. WTA is implemented by the biological hardware in order to select a location with maximum saliency, towards which to direct overt attention. In this paper we propose a gravitational model to describe the attentional shifts. Every single feature acts as an attractor and the shifts are the result of the joint effects of the attractors. In the current framework, the assumption of a single, centralized saliency map is no longer necessary, though still plausible. Quantitative results on two large image datasets show that this model predicts shifts more accurately than winner-take-all.
Asunto(s)
Atención/fisiología , Modelos Neurológicos , Percepción Visual/fisiología , Movimientos Oculares/fisiología , Humanos , Estimulación LuminosaRESUMEN
Humans are continuously exposed to a stream of visual data with a natural temporal structure. However, most successful computer vision algorithms work at image level, completely discarding the precious information carried by motion. In this paper, we claim that processing visual streams naturally leads to formulate the motion invariance principle, which enables the construction of a new theory of learning that originates from variational principles, just like in physics. Such principled approach is well suited for a discussion on a number of interesting questions that arise in vision, and it offers a well-posed computational scheme for the discovery of convolutional filters over the retina. Differently from traditional convolutional networks, which need massive supervision, the proposed theory offers a truly new scenario for the unsupervised processing of video signals, where features are extracted in a multi-layer architecture with motion invariance. While the theory enables the implementation of novel computer vision systems, it also sheds light on the role of information-based principles to drive possible biological solutions.
Asunto(s)
Aprendizaje Automático , Percepción de Movimiento/fisiología , Movimiento (Física) , Redes Neurales de la Computación , Estimulación Luminosa/métodos , Algoritmos , Animales , Bases de Datos Factuales , Humanos , Visión Ocular/fisiologíaRESUMEN
The understanding of the mechanisms behind focus of attention in a visual scene is a problem of great interest in visual perception and computer vision. In this paper, we describe a model of scanpath as a dynamic process which can be interpreted as a variational law somehow related to mechanics, where the focus of attention is subject to a gravitational field. The distributed virtual mass that drives eye movements is associated with the presence of details and motion in the video. Unlike most current models, the proposed approach does not estimate directly the saliency map, but the prediction of eye movements allows us to integrate over time the positions of interest. The process of inhibition-of-return is also supported in the same dynamic model with the purpose of simulating fixations and saccades. The differential equations of motion of the proposed model are numerically integrated to simulate scanpaths on both images and videos. Experimental results for the tasks of saliency and scanpath prediction on a wide collection of datasets are presented to support the theory. Top level performances are achieved especially in the prediction of scanpaths, which is the primary purpose of the proposed model.
Asunto(s)
Atención/fisiología , Movimientos Oculares/fisiología , Procesamiento de Imagen Asistido por Computador/métodos , Aprendizaje Automático , Algoritmos , Fijación Ocular/fisiología , Gravitación , Humanos , Modelos EstadísticosRESUMEN
This paper proposes a theory for understanding perceptual learning processes within the general framework of laws of nature. Artificial neural networks are regarded as systems whose connections are Lagrangian variables, namely, functions depending on time. They are used to minimize the cognitive action, an appropriate functional index that measures the agent interactions with the environment. The cognitive action contains a potential and a kinetic term that nicely resemble the classic formulation of regularization in machine learning. A special choice of the functional index, which leads to the fourth-order differential equations-Cognitive Action Laws (CAL)-exhibits a structure that mirrors classic formulation of machine learning. In particular, unlike the action of mechanics, the stationarity condition corresponds with the global minimum. Moreover, it is proven that typical asymptotic learning conditions on the weights can coexist with the initialization provided that the system dynamics is driven under a policy referred to as information overloading control. Finally, the theory is experimented for the problem of feature extraction in computer vision.
Asunto(s)
Cognición , Aprendizaje Automático , Redes Neurales de la Computación , Reconocimiento de Normas Patrones Automatizadas/métodos , Cognición/fisiología , Estimulación Luminosa/métodosRESUMEN
Eye movements are an essential part of human vision as they drive the fovea and, consequently, selective visual attention toward a region of interest in space. Free visual exploration is an inherently stochastic process depending on image statistics but also individual variability of cognitive and attentive state. We propose a theory of free visual exploration entirely formulated within the framework of physics and based on the general Principle of Least Action. Within this framework, differential laws describing eye movements emerge in accordance with bottom-up functional principles. In addition, we integrate top-down semantic information captured by deep convolutional neural networks pre-trained for the classification of common objects. To stress the model, we used a wide collection of images including basic features as well as high level semantic content. Results in a task of saliency prediction validate the theory.
Asunto(s)
Atención/fisiología , Movimientos Oculares/fisiología , Modelos Teóricos , Redes Neurales de la Computación , Percepción Visual/fisiología , HumanosRESUMEN
Malignant glomus tumors (MGTs) are exceptional but pose diagnostic and therapeutic challenges. Wide resection is the recommended treatment method, however, no data are available concerning adjuvant therapies. We present an exceptional case of extradigital deep-seated MGT of the forearm, with an exceptional bone infiltration. Despite being treated with wide resection, the patient had an optimal functional outcome, no functional loss, no motor or sensitive deficits and has returned to his full daily activity.