TPS1656 Background: Advances in big data analytics and artificial intelligence (AI) are enabling novel approaches to patient classification in oncology. While existing studies often correlate only a few data types, the DIPCAN Study (Digitalisation and Integral Management of Personalised Medicine in CANcer) seeks a comprehensive, integrated analysis combining phenotypic, clinical, pathological, radiomic, and genomic data from patients with metastatic cancer in Spain. DIPCAN aims to deepen insights into cancer’s multifactorial nature, driving personalized care and more precise therapeutic strategies. Methods: DIPCAN was initiated through a consortium comprising five technology and healthcare SMEs—Genomcore, Quibim, Pangaea Oncology, Artelnics, and Atrys Health—alongside Eurofins Megalab and the non-profit MD Anderson International Foundation Spain. Funding was secured via the Spanish Ministry of Economic Affairs and Digital Transformation under the EU-funded Recovery, Transformation, and Resilience Plan (R&D Missions Program in Artificial Intelligence, File No. MIA.2021.M02.0006). DIPCAN’s primary objective is to characterize and map clinical, phenotypic, genomic, and radiomic profiles of metastatic cancer patients across Spain. Secondary goals involve developing Big Data, AI, and machine learning tools to enable multidimensional analysis of these patients. Eligible patients are 18 years or older, have histologically confirmed metastatic solid tumors, a life expectancy exceeding three months, and available tumor material for histological and molecular analyses. Participants consent to undergo a comprehensive set of diagnostic and imaging procedures outlined in the study protocol. If recent tumor tissue (<3 years) is unavailable, patients may opt for a current biopsy or liquid biopsy. At no cost, participants receive consultations with oncology and drug development specialists, who document baseline characteristics and compile structured medical histories. Additional diagnostics include bloodwork emphasizing lipid metabolism, digital pathology, extensive NGS sequencing on tissue or blood, and a full-body MRI. All participants receive digital access to their data and a clinical report with tailored recommendations for their physicians. With ethics approval in place, DIPCAN has enrolled 1,500 patients since June 14, 2022. Data collection is ongoing, with anticipated advancements in AI-driven analysis aimed at refining precision oncology approaches for metastatic cancer in Spain. Clinical trial information: 2021.M02.0006 .
Material segmentation of satellites using multispectral imaging can support ground-based sensing and is often less sensitive to range and angular resolution. Many existing methods are validated primarily in simulation, leaving sim-to-real transfer under-evaluated. We introduce M2S2, comprising 12,960 synthetic and 3,024 hardware-in-the-loop (HIL) images across 10 spectral bands ($400-900 \text{nm}$), three satellite geometries, and up to eleven material classes. The dataset systematically varies elevation angles, material complexity, and lighting conditions, with synthetic wavelength-to-RGB approximations and true multispectral HIL captures. As a baseline, adapting Segment Anything (SAM) to multispectral inputs yields statistically significant gains over RGB on synthetic data, with advantages increasing with material count (up to $+4.64 {\%}$ macro recall), where 75% of non-two-class settings are significant at $\alpha=0.05$. HIL predictions exhibit uniformly high temporal consistency ($\overline{\text{mTC}}=0.9830$) with modest lighting effects (weak diffuse exceeding directional by 0.0024). These results suggest M2S2 is useful for characterizing sim-to-real challenges and for studying domain adaptation in satellite material segmentation. The dataset is available at https://huggingface.co/datasets/e-dealba/M2S2.
The objective of the HINMICO project is the development and optimization of manufacturing processes for the production of high-added value high quality multi-material micro-components, with the possibility of additional, functionalities, through more integrated, efficient and cost-effective process chains.
High performance polysulfone/γ-alumina biocompatible nanocomposites are reported for the first time and the effects of γ-alumina surface modification are explored. We show that some fatty acids chemisorb over the surface of γ-alumina forming nanosized self-assembled structures. These structures present thermal transitions at high temperatures, 100 °C higher than the melting temperatures of the pure acids, and are further shifted about 50 °C in the presence of polysulfone. The chemistry involved in the chemisorption is mild and green meeting the stringent bio sanitary protocols for biocompatible devices. It has been found that the self-assembled structures increase mechanical strength by about 20% despite the foreseeable lack of strong particle-matrix interactions, which manifests as small variations in both the glass transition temperature and the Young's modulus. Electron microscopy observation of fractured surfaces has revealed that some acids induce an extended region of influence around the nanoparticles and this fact has been used to explain the enhancement of mechanical strength.
