Stoppa, F., Ruiz de Austri, R., Vreeswijk, P., Bhattacharyya, S., Caron, S., Bloemen, S., et al. (2023). AutoSourceID-FeatureExtractor Optical image analysis using a two-step mean variance estimation network for feature estimation and uncertainty characterisation. Astron. Astrophys., 680, A108–14pp.
Abstract: Aims. In astronomy, machine learning has been successful in various tasks such as source localisation, classification, anomaly detection, and segmentation. However, feature regression remains an area with room for improvement. We aim to design a network that can accurately estimate sources' features and their uncertainties from single-band image cutouts, given the approximated locations of the sources provided by the previously developed code AutoSourceID-Light (ASID-L) or other external catalogues. This work serves as a proof of concept, showing the potential of machine learning in estimating astronomical features when trained on meticulously crafted synthetic images and subsequently applied to real astronomical data.Methods. The algorithm presented here, AutoSourceID-FeatureExtractor (ASID-FE), uses single-band cutouts of 32x32 pixels around the localised sources to estimate flux, sub-pixel centre coordinates, and their uncertainties. ASID-FE employs a two-step mean variance estimation (TS-MVE) approach to first estimate the features and then their uncertainties without the need for additional information, for example the point spread function (PSF). For this proof of concept, we generated a synthetic dataset comprising only point sources directly derived from real images, ensuring a controlled yet authentic testing environment.Results. We show that ASID-FE, trained on synthetic images derived from the MeerLICHT telescope, can predict more accurate features with respect to similar codes such as SourceExtractor and that the two-step method can estimate well-calibrated uncertainties that are better behaved compared to similar methods that use deep ensembles of simple MVE networks. Finally, we evaluate the model on real images from the MeerLICHT telescope and the Zwicky Transient Facility (ZTF) to test its transfer learning abilities.
|
Stoppa, F., Bhattacharyya, S., Ruiz de Austri, R., Vreeswijk, P., Caron, S., Zaharijas, G., et al. (2023). AutoSourceID-Classifier Star-galaxy classification using a convolutional neural network with spatial information. Astron. Astrophys., 680, A109–16pp.
Abstract: Aims. Traditional star-galaxy classification techniques often rely on feature estimation from catalogs, a process susceptible to introducing inaccuracies, thereby potentially jeopardizing the classification's reliability. Certain galaxies, especially those not manifesting as extended sources, can be misclassified when their shape parameters and flux solely drive the inference. We aim to create a robust and accurate classification network for identifying stars and galaxies directly from astronomical images.Methods. The AutoSourceID-Classifier (ASID-C) algorithm developed for this work uses 32x32 pixel single filter band source cutouts generated by the previously developed AutoSourceID-Light (ASID-L) code. By leveraging convolutional neural networks (CNN) and additional information about the source position within the full-field image, ASID-C aims to accurately classify all stars and galaxies within a survey. Subsequently, we employed a modified Platt scaling calibration for the output of the CNN, ensuring that the derived probabilities were effectively calibrated, delivering precise and reliable results.Results. We show that ASID-C, trained on MeerLICHT telescope images and using the Dark Energy Camera Legacy Survey (DECaLS) morphological classification, is a robust classifier and outperforms similar codes such as SourceExtractor. To facilitate a rigorous comparison, we also trained an eXtreme Gradient Boosting (XGBoost) model on tabular features extracted by SourceExtractor. While this XGBoost model approaches ASID-C in performance metrics, it does not offer the computational efficiency and reduced error propagation inherent in ASID-C's direct image-based classification approach. ASID-C excels in low signal-to-noise ratio and crowded scenarios, potentially aiding in transient host identification and advancing deep-sky astronomy.
|
Llosa, G., Barrio, J., Cabello, J., Crespo, A., Lacasta, C., Rafecas, M., et al. (2012). Detector characterization and first coincidence tests of a Compton telescope based on LaBr3 crystals and SiPMs. Nucl. Instrum. Methods Phys. Res. A, 695, 105–108.
Abstract: A Compton telescope for dose monitoring in hadron therapy consisting of several layers of continuous LaBr3 crystals coupled to silicon photomultiplier (SiPM) arrays is under development within the ENVISION project. In order to test the possibility of employing such detectors for the telescope, a detector head consisting of a continuous 16 mm x 18 mm x 5 mm LaBr3 crystal coupled to a SiPM array has been assembled and characterized, employing the SPIROC1 ASIC as readout electronics. The best energy resolution obtained at 511 key is 6.5% FWHM and the timing resolution is 3.1 ns FWHM. A position determination method for continuous crystals is being tested, with promising results. In addition, the detector has been operated in time coincidence with a second detector layer, to determine the coincidence capabilities of the system. The first tests are satisfactory, and encourage the development of larger detectors that will compose the telescope prototype.
|
Clinthorne, N., Brzezinski, K., Chesi, E., Cochran, E., Grkovski, M., Grosicar, B., et al. (2013). Silicon as an unconventional detector in positron emission tomography. Nucl. Instrum. Methods Phys. Res. A, 699, 216–220.
Abstract: Positron emission tomography (PET) is a widely used technique in medical imaging and in studying small animal models of human disease. In the conventional approach, the 511 keV annihilation photons emitted from a patient or small animal are detected by a ring of scintillators such as LYSO read out by arrays of photodetectors. Although this has been successful in achieving similar to 5 mm FWHM spatial resolution in human studies and similar to 1 mm resolution in dedicated small animal instruments, there is interest in significantly improving these figures. Silicon, although its stopping power is modest for 511 keV photons, offers a number of potential advantages over more conventional approaches including the potential for high intrinsic spatial resolution in 3D. To evaluate silicon in a variety of PET “magnifying glass” configurations, an instrument was constructed that consists of an outer partial-ring of PET scintillation detectors into which various arrangements of silicon detectors are inserted to emulate dual-ring or imaging probe geometries. Measurements using the test instrument demonstrated the capability of clearly resolving point sources of Na-22 having a 1.5 mm center-to-center spacing as well as the 1.2 mm rods of a F-18-filled resolution phantom. Although many challenges remain, silicon has potential to become the PET detector of choice when spatial resolution is the primary consideration. (C) 2012 Elsevier B.V. All rights reserved.
|
Magan, D. L. P., Caballero, L., Domingo-Pardo, C., Agramunt-Ros, J., Albiol, F., Casanovas, A., et al. (2016). First tests of the applicability of gamma-ray imaging for background discrimination in time-of-flight neutron capture measurements. Nucl. Instrum. Methods Phys. Res. A, 823, 107–119.
Abstract: In this work we explore for the first time the applicability of using gamma-ray imaging in neutron capture measurements to identify and suppress spatially localized background. For this aim, a pinhole gamma camera is assembled, tested and characterized in terms of energy and spatial performance. It consists of a monolithic CeBr3 scintillating crystal coupled to a position-sensitive photomultiplier and readout through an integrated circuit AMIC2GR. The pinhole collimator is a massive carven block of lead. A series of dedicated measurements with calibrated sources and with a neutron beam incident on a Au-197 sample have been carried out at n_TOF, achieving an enhancement of a factor of two in the signal-to-background ratio when selecting only those events coming from the direction of the sample.
|
