TY - JOUR A1 - Tsamos, Athanasios A1 - Evsevleev, Sergei A1 - Fioresi, R. A1 - Faglioni, F. A1 - Bruno, Giovanni T1 - Synthetic Data Generation for Automatic Segmentation of X-ray Computed Tomography Reconstructions of Complex Microstructures N2 - The greatest challenge when using deep convolutional neural networks (DCNNs) for automatic segmentation of microstructural X-ray computed tomography (XCT) data is the acquisition of sufficient and relevant data to train the working network. Traditionally, these have been attained by manually annotating a few slices for 2D DCNNs. However, complex multiphase microstructures would presumably be better segmented with 3D networks. However, manual segmentation labeling for 3D problems is prohibitive. In this work, we introduce a method for generating synthetic XCT data for a challenging six-phase Al–Si alloy composite reinforced with ceramic fibers and particles. Moreover, we propose certain data augmentations (brightness, contrast, noise, and blur), a special in-house designed deep convolutional neural network (Triple UNet), and a multi-view forwarding strategy to promote generalized learning from synthetic data and therefore achieve successful segmentations. We obtain an overall Dice score of 0.77. Lastly, we prove the detrimental effects of artifacts in the XCT data on achieving accurate segmentations when synthetic data are employed for training the DCNNs. The methods presented in this work are applicable to other materials and imaging techniques as well. Successful segmentation coupled with neural networks trained with synthetic data will accelerate scientific output. KW - Automatic segmentation KW - 3D deep convolutional neural network (3D DCNN) KW - Dice score KW - Metal matrix composite (MMC) KW - Modified U-Net architectures KW - Multi-phase materials PY - 2023 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-571243 DO - https://doi.org/10.3390/jimaging9020022 VL - 9 IS - 2 SP - 1 EP - 23 PB - MDPI AN - OPUS4-57124 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Roveda, Ilaria A1 - Serrano-Munoz, Itziar A1 - Mishurova, Tatiana A1 - Madia, Mauro A1 - Pirling, T. A1 - Evans, Alexander A1 - Klaus, M. A1 - Haubrich, J. A1 - Requena, G. A1 - Bruno, Giovanni T1 - Influence of a 265 °C heat treatment on the residual stress state of a PBF-LB/M AlSi10Mg alloy N2 - Laser Powder Bed Fusion (PBF-LB/M) additive manufacturing (AM) induces high magnitude residual stress (RS) in structures due to the extremely heterogeneous cooling and heating rates. As the RS can be deleterious to the fatigue resistance of engineering components, great efforts are focused on understanding their generation and evolution after post-process heat treatments. In this study, one of the few of its kind, the RS relaxation induced in an as-built PBF-LB/M AlSi10Mg material by a low-temperature heat treatment (265 °C for 1 h) is studied by means of X-ray and neutron diffraction. Since the specimens are manufactured using a baseplate heated up to 200 °C, low RS are found in the as-built condition. After heat treatment a redistribution of the RS is observed, while their magnitude remains constant. It is proposed that the redistribution is induced by a repartition of stresses between the a-aluminium matrix and the silicon phase, as the morphology of the silicon phase is affected by the heat treatment. A considerable scatter is observed in the neutron diffraction RS profiles, which is principally correlated to the presence (or absence) of pockets of porosity developed at the borders of the chessboard pattern. KW - Neutron diffraction KW - Additive manufacturing PY - 2022 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-565115 DO - https://doi.org/10.1007/s10853-022-07997-w SN - 1573-4803 VL - 57 SP - 22082 EP - 22098 PB - Springer Science + Business Media CY - Dordrecht AN - OPUS4-56511 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Lapenna, Michela A1 - Faglioni, Francesco A1 - Chand, Keerthana A1 - Hejazi, Bardia A1 - Fioresi, Rita A1 - Bruno, Giovanni T1 - Chamfer distance for non-linear registration of Triply Periodic Minimal Surface lattices N2 - We present a 3D image registration technique for non-linear deformation estimation in Additive Manufacturing processes. The methodology involves comparing X-ray Computed Tomography (XCT) data with Computer Aided Design (CAD) models for Triply Periodic Minimal Surface (TPMS) lattices and employs the Chamfer distance to refine mesh non-linear deformations. KW - X-ray Computed tomography KW - Defects KW - Machine Learning KW - Digital Twin KW - Registration PY - 2025 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-637367 DO - https://doi.org/10.1016/j.addlet.2025.100299 SN - 2772-3690 VL - 14 SP - 1 EP - 8 PB - Elsevier B.V. AN - OPUS4-63736 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Rieder, Philipp A1 - Petrich, Lukas A1 - Serrano-Munoz, Itziar A1 - Mouiya, Mossaab A1 - Markötter, Henning A1 - Huger, Marc A1 - Bruno, Giovanni A1 - Schmidt, Volker T1 - Statistical analysis of grains and pores within polycrystalline Al2TiO5 ceramics, based on X-ray computed tomography N2 - Advanced statistical image analysis workflows were developed to segment and quantitatively evaluate 2D electron-backscatter diffraction (EBSD) maps and 3D synchrotron X-ray computed tomography (SXCT) volumes of a polycrystalline Al2TiO5 refractory composite that contains microcracks and pores. Several size, shape, and further geometric descriptors were determined for both the solid phase (Al2TiO5 grains) and the pore space. The resulting pore-size distribution is distinctly bimodal: coarse pores (tens to hundreds of micrometers), traced to incomplete powder compaction, coexist with fine pores generated during sintering. The two pore populations appear to be correlated with grain growth and crystallographic orientation in different ways. Finally, the descriptors obtained from the 2D EBSD and 3D SXCT data sets are internally consistent but complementary, highlighting the value of characterizations based on EBSD and SXCT in the microstructural study of refractory ceramics. KW - Synchrotron X-ray computed tomography KW - Electron-backscatter diffraction KW - Aluminum titanate KW - Morphological reconstruction KW - Image Segmentation KW - Statistical image analysis PY - 2025 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-643918 DO - https://doi.org/10.1016/j.matchar.2025.115602 SN - 1044-5803 VL - 229 SP - 1 EP - 13 PB - Elsevier Inc. AN - OPUS4-64391 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Buljak, Vladimir A1 - Serrano-Munoz, Itziar A1 - Kupsch, Andreas A1 - Müller, Bernd R. A1 - Prasek, Marko A1 - Contillo, Adriano A1 - Mouiya, Mossaab A1 - Huger, Marc A1 - Bruno, Giovanni T1 - On the closure of thermally induced micro-cracks in aluminum titanate ceramics N2 - Aluminum Titanate (AT) refractory ceramics (as some other ceramic composites) are prone to microcracking, due to the thermal expansion anisotropy of AT and to the mismatch with the thermal expansion of the constituents. Such microcracks cause the room temperature Young's modulus to be only a fraction of that of the non-microcracked material. As a function of temperature, the Young's modulus increases non-linearly. Such increase suggests that microcracks close or even heal at high temperatures. Upon cooling, thermal stress accumulates again, and microcracks re-open. This cycle is fully reversible. While confirming the hysteretic behavior of the Young's modulus, we observe that the amount of microcracks (as determined by in-situ Synchrotron X-ray refraction radiography) decreases linearly upon heating. The apparent mismatch between the Young's modulus and the microcrack content dependence on temperature is explained by a simple FEM model. Such model employs cohesive elements upon cooling, in order to estimate the amount of initial microcracks. On purpose, the model does not include healing upon heating and only allows crack closure. It predicts that crack closure continuously occurs upon heating, thereby qualitatively reproducing the nearly linear dependence of the X-ray refraction signal. It is therefore concluded that the sudden and non-linear increase of Young's modulus with temperature is mainly caused by crack healing. Such finding agrees with previous work and paves the road to a more systematic separation of crack closure and healing in flexible ceramics. KW - X-ray computed tomography KW - X-ray refraction radiography KW - In-situ imaging KW - BAMline KW - Micromechanics KW - BESSY II PY - 2025 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-645082 DO - https://doi.org/10.1016/j.ceramint.2025.09.237 SN - 0272-8842 VL - 51 IS - 27 SP - 55141 EP - 55152 PB - Elsevier CY - Amsterdam AN - OPUS4-64508 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Bruno, Giovanni A1 - Lapenna, M. A1 - Faglioni, F. A1 - Fioresi, R. T1 - Temperature-based pruning for input features in Graph Neural Networks N2 - In the presentwork,we employ the concept of neural network temperature to prune unimportant features in input to aGraph Neural Network (GNN) architecture. In benchmark datasets for node and graph property prediction, each node comes equipped with a vector of numerous features. It is paramount to understand which information is actually necessary and which can be discarded, both for efficiency and explainability. The temperature is linked to the gradient activity due to the loss function minimization and leads to pruning of weight structures associated with small gradients. This study is done on different GNN architectures, one for node classification and another one for link prediction, and several benchmark datasets are employed.We compare the results with similar experiments previously conducted on the filters of Convolutional Neural Networks. Although still at the proof-of-concept stage, our temperature-based pruning technique stands as a promising alternative to state-of-the-art magnitude-based pruning techniques. KW - Temperature-based pruning KW - Graph Neural Networks PY - 2025 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-640886 VL - 140 SP - 1 EP - 20 PB - Springer AN - OPUS4-64088 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Gunnerek, R. A1 - Soundarapandiyan, G. A1 - Mishurova, T. A1 - Schröder, J. A1 - Bruno, Giovanni A1 - Boykin, J. A1 - Diaz, A. A1 - Klement, U. A1 - Hryha, E.ON T1 - Chemical mechanical polishing of powder bed fusion – laser beam processed 316 L stainless steel N2 - Additive manufacturing via powder bed fusion – laser beam (PBF-LB) enables the fabrication of complex geometries but suffers from inherently rough surfaces and surface tensile residual stresses, both of which can compromise structural integrity, particularly under fatigue loading. To address these limitations, this study investigates chemical mechanical polishing (CMP) as a surface finishing method for improving surface quality and modifying the residual stress state in PBF-LB 316 L stainless steel. The work uniquely examines how scan rotation (0◦ vs. 67◦ rotation) and contour parameters influence CMP effectiveness in material removal, surface smoothing, and subsurface stress redistribution. With a targeted material removal of 110 μm, CMP reduced surface roughness (Sa) by up to 94 %, achieving values as low as 0.7 μm. Microstructural analysis revealed no grain refinement but identified a thin, plastically deformed surface layer. This plastic deformation resulted in the transformation of tensile surface stresses (340 MPa) into beneficial compressive stresses (􀀀 400 MPa), as confirmed by synchrotron X-ray diffraction, which also showed a shift toward isotropic strain distribution. Further, these findings demonstrate that the initial scan strategy influences CMP performance and that CMP can enhance both surface integrity and mechanical reliability without altering the underlying microstructure. This study advances the understanding of how process induced microstructure and surface features affect CMP outcomes, enabling more informed design of post-processing strategies for improved surface integrity and mechanical performance in additively manufactured metals. KW - Residual stress KW - Additive manufacturing KW - Chemical mechanical polishing KW - As-built microstructure KW - Surface roughness KW - Surface finishing KW - Material removal PY - 2025 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-640522 DO - https://doi.org/10.1016/j.jmatprotec.2025.119055 SN - 0924-0136/ VL - 345 SP - 1 EP - 12 PB - Elsevier B.V. AN - OPUS4-64052 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CHAP A1 - Hejazi, Bardia A1 - Chand, Keerthana A1 - Fritsch, Tobias A1 - Bruno, Giovanni T1 - D-CNN and VQ-VAE Autoencoders for Compression and Denoising of Industrial X-Ray Computed Tomography Images N2 - The ever-growing volume of data in imaging sciences stemming from advancements in imaging technologies, necessitates efficient and reliable storage solutions for such large datasets. This study investigates the compression of industrial X-ray computed tomography (XCT) data using deep learning autoencoders and examines how these compression algorithms affect the quality of the recovered data. Two network architectures with different compression rates were used, a deep convolution neural network (D-CNN) and a vector quantized variational autoencoder (VQ-VAE). The XCT data used was from a sandstone sample with a complex internal pore network as a good test case for the importance of feature preservation. The quality of the decoded images obtained from the two different deep learning architectures with different compression rates were quantified and compared to the original input data. In addition, to improve image decoding quality metrics, we introduced a metric sensitive to edge preservation, which is crucial for three-dimensional data analysis. We showed that different architectures and compression rates are required depending on the specific characteristics needed to be preserved for later analysis. The findings presented here can aid scientists in determining the requirements and strategies needed for appropriate data storage and analysis. T2 - 28th European Conference on Artificial Intelligence – Including 14th Conference on Prestigious Applications of Intelligent Systems (PAIS 2025) CY - Bologna, Italy DA - 25.10.2025 KW - Data Compression KW - Deep Learning KW - X-ray Computed Tomography PY - 2025 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-644758 UR - https://ebooks.iospress.nl/doi/10.3233/FAIA251480 DO - https://doi.org/10.3233/FAIA251480 SN - 0922-6389 SP - 1 EP - 8 PB - IOS Press AN - OPUS4-64475 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Pfaff, Jonas A1 - Schopferer, Sebastian A1 - Markötter, Henning A1 - Rack, Alexander A1 - Bruno, Giovanni A1 - Schmidt, Anita A1 - Tichter, Tim A1 - Böttcher, Nils T1 - High-speed synchrotron radiography of nail penetration-induced thermal runaway: Understanding the explosive behavior of commercial sodium-ion batteries with NFM cathode N2 - The dynamics of mechanically initiated thermal runaway (TR) events in cylindrical 18650 cells with NFM (Na(Ni1/3Fe1/3Mn1/3)O2), LFP (LiFePO4), and NMC532 (LiNi1/2Mn1/3Co1/5O2) cathode chemistries were investigated using high-speed synchrotron X-ray imaging. Structural similarity index measures (SSIM) were employed to identify and track rapid structural changes. In this manner, thermal decompositions and internal propagation dynamics, influencing the safety mechanisms of the cells, were studied. This lead to two major findings: (I) Among NFM, LFP, and NMC532 cells, the TR-characteristics differ significantly in temperature and internal propagation speed. Internal safety mechanisms appear, however, visually similar. Among all samples, LFP cells exhibit higher safety performance concerning the initiation of TR by nail penetration and the progression of TR. (II) The NFM cells used in this study displayed an almost explosive TR. This finding appears counterintuitive on a first glance, since sodium-ion batteries are usually considered safe. High-speed imaging revealed that the explosive TR is not necessarily caused by the thermochemical decomposition reactions, but rather by a failure of the venting mechanism. This results in a significant pressure buildup within the cell upon TR initiation and eventually a severely violent TR. These results underline that battery safety depends on many factors and not solely on optimized cell chemistries or materials. KW - High-speed X-ray radiography KW - Synchrotron X-rays KW - 18650 KW - SIB KW - Na-NFM KW - Structure similarity index measure PY - 2025 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-642649 DO - https://doi.org/10.1016/j.powera.2025.100188 SN - 2666-2485 VL - 36 SP - 1 EP - 7 PB - Elsevier Ltd. AN - OPUS4-64264 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Erdmann, Maren A1 - Kupsch, Andreas A1 - Müller, Bernd R. A1 - Hentschel, Manfred P. A1 - Niebergall, Ute A1 - Böhning, Martin A1 - Bruno, Giovanni ED - Erdmann, Maren T1 - Diesel-induced transparency of plastically deformed high-density polyethylene N2 - High-density polyethylene becomes optically transparent during tensile drawing when previously saturated with diesel fuel. This unusual phenomenon is investigated as it might allow conclusions with respect to the material behavior. Microscopy, differential scanning calorimetry, density measurements are applied together with two scanning X-ray scattering techniques: wide angle X-ray scattering (WAXS) and X-ray refraction, able to extract the spatially resolved crystal orientation and internal surface, respectively. The sorbed diesel softens the material and significantly alters the yielding characteristics. Although the crystallinity among stretched regions is similar, a virgin reference sample exhibits strain whitening during stretching, while the diesel-saturated sample becomes transparent. The WAXS results reveal a pronounced fiber texture in the tensile direction in the stretched region and an isotropic orientation in the unstretched region. This texture implies the formation of fibrils in the stretched region, while spherulites remain intact in the unstretched parts of the specimens. X-ray refraction reveals a preferred orientation of internal surfaces along the tensile direction in the stretched region of virgin samples, while the sample stretched in the diesel-saturated state shows no internal surfaces at all. Besides from stretching saturated samples, optical transparency is also obtained from sorbing samples in diesel after stretching. KW - PE-HD Sorption KW - Cavitation KW - Diesel Fuel KW - X-ray refraction KW - WAXS KW - Internal Surfaces KW - Crystal Texture PY - 2019 DO - https://doi.org/10.1007/s10853-019-03700-8 SN - 1573-4803 SN - 0022-2461 VL - 54 IS - 17 SP - 11739 EP - 11755 PB - Springer US CY - US AN - OPUS4-48226 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER -