RESUMEN
Recent advances in depth measurement and its utilization have made point cloud processing more critical. Additionally, the human head is essential for communication, and its three-dimensional data are expected to be utilized in this regard. However, a single RGB-Depth (RGBD) camera is prone to occlusion and depth measurement failure for dark hair colors such as black hair. Recently, point cloud completion, where an entire point cloud is estimated and generated from a partial point cloud, has been studied, but only the shape is learned, rather than the completion of colored point clouds. Thus, this paper proposes a machine learning-based completion method for colored point clouds with XYZ location information and the International Commission on Illumination (CIE) LAB (L*a*b*) color information. The proposed method uses the color difference between point clouds based on the Chamfer Distance (CD) or Earth Mover's Distance (EMD) of point cloud shape evaluation as a color loss. In addition, an adversarial loss to L*a*b*-Depth images rendered from the output point cloud can improve the visual quality. The experiments examined networks trained using a colored point cloud dataset created by combining two 3D datasets: hairstyles and faces. Experimental results show that using the adversarial loss with the colored point cloud renderer in the proposed method improves the image domain's evaluation.
RESUMEN
For visually handicapped people, a mental support is important in their independent daily life and participation in a society. It is expected to develop a system which can recognize colors and patterns on clothes so that they can go out with less concerns. We have worked on a basic study into such a system, and developed a prototype system which can stably recognize colors and patterns and immediately provide these information in voice, when a user faces it to clothes. In the results of evaluation experiments it is shown that the prototype system is superior to the system in the basic study at the accuracy rate for the recognition of color and pattern.
Asunto(s)
Personas con Daño Visual/psicología , Voz/fisiología , Algoritmos , Vestuario , Color , Equipos de Comunicación para Personas con Discapacidad , Humanos , Interfaz Usuario-ComputadorRESUMEN
PURPOSE: [7-methyl-(11)C]-(E)-8-(3,4,5-trimethoxystyryl)-1,3,7-trimethylxanthine ([(11)C]TMSX) is a positron-emitting adenosine A(2A) receptor (A2AR) antagonist for visualisation of A2AR distribution by positron emission tomography (PET). The aims of this paper were to use a kinetic model to analyse the behaviour of [(11)C]TMSX in the brain and to examine the applicability of the Logan plot. We also studied the applicability of a simplified Logan plot by omitting metabolite correction and arterial blood sampling. METHODS: The centrum semiovale was used as a reference region on the basis of a post-mortem study showing that it has a negligibly low density of A2ARs. Compartmental analysis was performed in five normal subjects. Parametric images of A2AR binding potential (BP) were also generated using a Logan plot with or without metabolite correction and with or without arterial blood sampling. To omit arterial blood sampling, we applied a method to extract the plasma-related information using independent component analysis (EPICA). RESULTS: The estimated K (1)/k (2) was confirmed to be common in the centrum semiovale and main cortices. The three-compartment model was well fitted to the other regions using the fixed value of K (1)/k (2) estimated from the centrum semiovale. The estimated BPs using the Logan plot matched those derived from compartment analysis. Without the metabolite correction, the estimate of BP underestimated the true value by 5%. The estimated BPs agreed regardless of arterial blood sampling. CONCLUSION: A three-compartment model with a reference region, the centrum semiovale, describes the kinetic behaviour of [(11)C]TMSX PET images. A2ARs in the human brain can be visualised as a BP image using [(11)C]TMSX PET without arterial blood sampling.
Asunto(s)
Encéfalo/diagnóstico por imagen , Encéfalo/patología , Tomografía de Emisión de Positrones/métodos , Receptor de Adenosina A2A/metabolismo , Xantinas/farmacocinética , Adulto , Radioisótopos de Carbono/farmacocinética , Interpretación Estadística de Datos , Humanos , Procesamiento de Imagen Asistido por Computador , Cinética , Análisis de los Mínimos Cuadrados , Masculino , Modelos Biológicos , Sensibilidad y Especificidad , Factores de TiempoRESUMEN
We have previously proposed a statistical method for extracting a plasma time-activity curve (pTAC) from dynamic PET images, named EPICA, for kinetic analysis of cerebral glucose metabolism. We assumed that the dynamic PET images consist of a blood-related component and a tissue-related component which are spatially independent in a statistical sense. The aim of this study is to investigate the utility of EPICA in imaging total distribution volume (DVt) and binding potential (BP) with Logan plots in a neuroreceptor mapping study. We applied EPICA to dynamic [(11)C]MPDX PET images in 25 subjects, including healthy subjects and patients with brain diseases, and validated the estimated pTACs. [11C]MPDX is a newly developed radiopharmaceutical for mapping cerebral adenosine A1 receptors. EPICA successfully extracted pTAC for all 25 subjects. Parametric images of DVts were estimated by applying Logan plots with the EPICA-estimated pTAC and then used to define a reference region. The BPs estimated using EPICA were evaluated in 18 subjects by ROI-based comparison with those obtained using the nonlinear least squares method (NLSM). The calculated BPs were identical to the estimates using NLSM in each subject. We conclude that EPICA is a promising technique that generates parametric images of DVt and BP in neuroreceptor mapping without requiring arterial blood sampling.