Thesis Overview

Brain pathology analysis through medical imaging is critical for clinical diagnosis, treatment planning, and patient outcome prediction. Computed tomography (CT) and magnetic resonance imaging (MRI) have become essential imaging modalities for assessing two major classes of brain pathologies: intracranial hemorrhage (ICH) and brain tumors. However, manual interpretation and analysis of these imaging studies are time-consuming, labor-intensive, and susceptible to intra- and interobserver variability. In recent years, artificial intelligence-based methods have shown great potential in automating the detection, segmentation, and quantification of brain lesions, thereby supporting clinical workflows and improving diagnostic accuracy. Nevertheless, these methods typically rely on large amounts of annotated data to achieve high performance. In medical imaging, obtaining such annotations is particularly challenging due to the need for expert knowledge, as well as the significant time and economic costs associated with manual labeling. As a result, while vast amounts of medical imaging data are routinely acquired in clinical practice, only a small fraction is accompanied by high-quality annotations. This imbalance highlights the need for approaches that can effectively leverage limited labeled data, motivating the development of learning strategies under limited supervision.

The main goal of this PhD thesis is to develop robust deep learning methodologies for neuroimaging analysis under limited supervision. To address annotation scarcity, we explore several complementary strategies. In the first stage, we propose a framework for hematoma expansion (HE) prediction in spontaneous intracerebral hemorrhage (ICH) using baseline CT scans. This approach combines a classification model with synthetic CT image generation to enhance robustness and alleviate data limitations. Our results show that integrating synthetic data with conventional augmentation improves both predictive performance and generalization. In the second stage, we develop semi-supervised deep learning frameworks that leverage both labeled and unlabeled data for binary brain lesion segmentation in brain tumors and spontaneous ICHs. We introduce a novel method that combines feature perturbation with mutual learning, achieving significant improvements in low-annotation settings (10–20% labeled data). In the third stage, we extend semi-supervised learning to multi-class hemorrhage segmentation, a more complex task requiring the differentiation of multiple hemorrhage subtypes. We conduct a comprehensive benchmark of established semi-supervised methods within a unified nnUNet framework, providing a systematic evaluation of different strategies for this setting.

All proposed methodologies are evaluated on both institutional datasets from collaborating hospitals and public international benchmarks, enabling objective comparison with state-of-the-art approaches. Notably, this work was carried out in close collaboration with the medical team at Hospital Dr. Josep Trueta, with clinical expertise integrated throughout the research process from problem formulation and data acquisition to result interpretation and validation. By combining deep learning techniques with clinical insight, this thesis aims to deliver practical tools that support clinical decision-making and ultimately improve patient care in brain tumors and ICHs. Emphasis is placed on reproducibility and clinical applicability, with all methods and experimental pipelines made publicly available to facilitate future research and the development of reliable automated decision-support systems.

The post Doctoral Thesis: Deep Learning Methods with Limited Supervision for Brain Pathology Analysis in Neuroimaging appeared first on Vicorob.

The delegation from the Medical Imaging Lab included Dr. Robert Martí, who served as a member of the workshop committee and chaired a scientific session, alongside PhD candidates Hadeel Awwad and Enric Sena. The workshop provided a valuable platform for the team to engage with international imaging peers, including newly onboarding researchers like Enric.

The work presented falls under the framework of the IMPACT project, which leverages deep learning to drive innovation in image analysis, optimize lesion detection, and advance treatment response prediction to promote personalized healthcare outcomes.

PhD candidate Hadeel Awwad presented a poster detailing preliminary results on predicting pathologic complete response (pCR) from Breast MRI using the multi-centre MAMA-MIA dataset. To find the configuration with the highest discriminative power, the study conducted a systematic evaluation of 78 different radiomics configurations spanning spatial regions—specifically comparing the primary tumour and a peritumoural ring—as well as temporal representations and mathematical transforms.

The preliminary findings revealed that simple, dynamic, first-order peritumoural signals consistently achieved the highest classification accuracy, outperforming static or purely intratumoural features. These results justify integrating these localized temporal variables into future comprehensive models to improve treatment response prediction.

Beyond the scientific presentations, the workshop served as an excellent venue for strengthening ties within the international breast imaging community. The team took advantage of the event to establish new collaborations and lay the groundwork for future joint research initiatives with European and international partners.

This work has been funded by the IMPACT project from the Ministerio de Ciencia e Innovación y Universidades of Spain, and the PhD grant IFUdG2024 from the University of Girona.

The post Latest breast MRI research from ViCOROB featured at IWBI 2026 appeared first on Vicorob.

Thesis Overview

This PhD thesis focuses on the development of artificial intelligence methods based on deep learning for the analysis of brain images and, ultimately, to help improve the diagnosis and treatment of common neurological diseases such as stroke and meningioma.

Stroke is one of the leading causes of death and disability worldwide. It can be either ischemic or hemorrhagic, and in both cases the diagnosis relies on brain imaging examinations, most commonly computed tomography (CT). Meningioma, on the other hand, is the most common primary brain tumor. It is often detected and monitored through periodic magnetic resonance imaging (MRI) examinations, and when treatment is required, especially radiotherapy, treatment planning is based on these images to accurately define the area to be irradiated. Therefore, medical imaging is a key tool throughout all stages of the clinical management of these diseases.

In this context, this thesis explores how deep learning can make different steps of the clinical workflow more efficient and accurate. First, a method is proposed to predict the possible evolution of intracerebral hemorrhage in patients with hemorrhagic stroke using only the initial available CT scan. One of the main challenges is the limited availability of clinical data, and therefore a strategy is also proposed to generate synthetic examples simulating possible lesion evolutions. These synthetic images are used to train the model and improve its predictive capability, achieving competitive results compared with state-of-the-art methods.

Second, a system is developed to detect large vessel occlusions, one of the main causes of ischemic stroke, from CT angiography images, and to identify the affected vascular segment. The method focuses on a specific anatomical region, the Circle of Willis, which contains the most relevant vessels. This allows the analysis to be simplified, the process to be accelerated, and a high level of reliability to be maintained. The results have been validated using real hospital data and independent test sets, demonstrating good generalization capability.

Third, a method is presented for the segmentation of meningioma tumors in MRI images used for radiotherapy planning. The main contribution is a richer representation of the tumor, which considers not only the lesion itself but also its transition zone. This additional information improves segmentation accuracy and achieves better results, even in comparison with other recent approaches. The method was evaluated and validated in an international challenge, where it achieved first place.

Overall, this thesis demonstrates how artificial intelligence can contribute to making brain image analysis more automated, accurate, and efficient. An important part of the work was carried out in collaboration with the medical team at Hospital Dr. Josep Trueta, integrating real clinical knowledge throughout all stages of the project. The ultimate goal is to provide tools that can have a direct impact on clinical practice and contribute to improving the care of patients with stroke and brain tumors.

The post Doctoral Thesis: Deep Learning for Clinical Decision Support in Brain Tumors and Stroke appeared first on Vicorob.

The project aims to develop new machine learning approaches for the Girona1000 AUV, that integrates side-scan sonar with a dual-camera optical system, to enhance underwater mapping and blue carbon inventory, providing valuable information for environmental monitoring and conservation efforts. By generating and analysing side-scan sonar maps in real time, the robot can interpret the underwater environment as it explores the seafloor and autonomously adapt its mission based on the information it acquires, improving the efficiency and quality of underwater inspection and mapping operations.

The work was presented by Dr. Nuno Gracias, together with PhD candidates Hayat Rajani and Valerio Franchi from CIRS. The poster introduce the first underwater stereo event-camera setup developed within the project. The system uses an AMD Kria KV260 to preprocess the large volume of event data before feeding it to a SpiNNode neuromorphic board, built around the SpiNNaker2 chip, which runs spiking neural networks for classification. The high temporal resolution and high dynamic range of event cameras cope with the rapid motion and uneven illumination underwater, while their low power consumption supports the long survey missions that broad blue carbon inventory requires.

Valerio and Hayat also took the opportunity to present ongoing work from two other projects. One of the posters introduced a multi-robot system for total body photography at dermatoscopic image quality, a line of research begun under the iToBoS project and continued since its conclusion. The work-in-progress framework uses four UR10 manipulators with end-effector cameras, combining view planning and multi-robot scheduling with a real-time mole tracker in a visual servoing loop, allowing it to adapt to individual patient anatomies and capture consistent high-resolution images while reducing examination time and operator dependency. The other poster reported preliminary results from the ongoing tech transfer project OculaRobot an autonomous framework for precision intravitreal injection, one of the highest-volume medical procedures worldwide. The system pairs a 6-DOF collaborative manipulator with a trifocal stereoscopic vision rig for real-time 3D tracking of the eye and needle, using visual servoing and force feedback to target sub-millimetre positioning accuracy. Both posters were presented at the well-regarded workshop on robot-assisted medical imaging, RAMI .

Additionally, Hayat presented a poster from the work carried out at Heriot-Watt University, introducing a controlled benchmark designed to systematically evaluate modern 3D reconstruction methods under variations in medium and lighting. The benchmark comprises 13 datasets spanning two media (air and water) and three lighting conditions (ambient, artificial, and mixed), with additional variations in motion type, scanning pattern, and initialization trajectory, resulting in a diverse set of sequences.

Funded by the Agencia Estatal de Investigación (AEI) and the NextGenerationEU programme, the IURBI project is now entering its final stage. The technologies developed within the project contribute to the advancement of more intelligent and autonomous underwater robotic systems, with potential applications in marine exploration, environmental monitoring and the conservation of marine ecosystems, reinforcing the role of technological innovation in addressing scientific and societal challenges.

The post ViCOROB presents latest IURBI advances and research innovations at ICRA 2026 appeared first on Vicorob.

As part of the conference programme, the presentation entitled “DIVE INTO LEARNING Project: Inclusive STEAM Education with Low-Cost Underwater Robots” was delivered to showcase this Erasmus+ project, which aims to promote more inclusive STEAM education through the use of low-cost underwater robots and project-based learning methodologies.

The project seeks to foster social inclusion, support teachers, and enhance the participation of students from underrepresented and minority groups through innovative educational activities. Within this framework, DIVE Into Learning is developing a curriculum based on underwater robotics and marine sciences to bring technology and research into new educational contexts while ensuring that all learners can successfully acquire knowledge and skills in these disciplines.

Participation in CUIEET, a conference that brings together engineering schools from across Spain, provided an opportunity to increase the project’s visibility at the national level and to share this educational approach with the academic community interested in educational innovation and inclusive STEAM methodologies.

The post Vicorob presents the DIVE Into Learning project at “Congreso Universitario de Innovación Educativa en las Enseñanzas Técnicas” appeared first on Vicorob.

La sessió va apropar al públic alguns dels avenços més recents en l’estudi del cervell humà i el paper que hi juga actualment la intel·ligència artificial. Durant la conferència, Casamitjana va explicar com les noves eines computacionals permeten connectar diferents nivells d’estudi del cervell, des de l’escala microscòpica de les cèl·lules fins a l’anàlisi macroscòpica dels sistemes cerebrals i la seva interrelació.

La conferència també va presentar el projecte NextBrain, una iniciativa de recerca que utilitza tècniques d’intel·ligència artificial per oferir un mapa complet de l’arquitectura cel·lular del cervell sense precedents. Les tècniques desenvolupades permeten una anàlisi ràpida, precisa i accessible d’imatges cerebrals en individus vius. Aquest enfoc permet identificar patrons cerebrals complexos, facilitant l’estudi del desenvolupament i l’envelliment cerebral, així com en l’impacte de diverses malalties neurodegeneratives. Totes les eines del projecte NextBrain han estat publicades lliurement a la comunitat científica per a futurs estudis: https://github-pages.ucl.ac.uk/NextBrain

Veiem, doncs, com la intel·ligència artificial esdevé una tecnologia crucial en l’estudi del cervell, ja que permet examinar grans quantitats de dades amb patrons complexes. Obre les portes a estudis fins ara inimaginables, integrant dades de múltiples fonts que anticipin la detecció de malalties cerebrals. Ens permetrà, per exemple, poder dissenyar protocols i intervencions personalitzades i, fins i tot, crear bessons digitals que permetin simular la evolució dels pacients i l’impacte dels tractaments.

El cicle “Últimes fronteres. La ciència al segle XXI”, organitzat per la Casa de Cultura de Girona, té com a objectiu apropar la recerca científica i tecnològica més actual a la ciutadania. La sessió es va gravar en vídeo i estarà disponible al canal de Youtube de la Casa de Cultura de Girona.

The post La recerca en IA aplicada al cervell humà centra una conferència d’Adrià Casamitjana appeared first on Vicorob.

The applications presented span high-impact domains such as ecology (biodiversity estimation, tree species census, and biomass assessment) and cultural heritage preservation through the development of large-scale databases of historical documents.

The visit also included working meetings with the obstetrics team at Dr. Josep Trueta University Hospital, involving Dr. Anna Maroto and Dr. Clara Martínez, to explore ongoing research lines in fetal ultrasound imaging and identify new opportunities for clinical collaboration.

This activity is part of a long-standing collaboration between the two researchers, initiated in 2019 and consolidated through joint publications and competitive research proposals. In this context, it is also framed within the Ramón y Cajal programme awarded to Mariano Cabezas (RYC2023-042706-I), one of Spain’s flagship initiatives for the recruitment and consolidation of research talent, which supports his upcoming research stay in Japan and enables the expansion of these joint research lines. In parallel, Yago Díez’s involvement is linked to the J-PEAKS programme (2026 J-PEAKS Accelerated International Network Construction Support Program), aimed at fostering international research networks.

The post International collaboration in Computer Vision, Deep Learning and Medical Imaging appeared first on Vicorob.

A Day of Collaboration and Learning

The day was thoughtfully structured to combine strategic planning with practical experience:

In the morning, partners gathered at UdiGitalEdu, the research group of the University of Girona coordinating the project, to:

• Establish project protocols ensuring smooth coordination and collaboration
• Plan dissemination activities to raise awareness and engage target groups, focusing on secondary multicultural schools
• Co-design tools for collecting data to perform a thorough current situation analysis, ensuring that project resources are based on solid evidence and informed decisions

In the afternoon, the group moved to the CIRS (Underwater Vision and Robotics Research Centre), where partners:

• Participated in capacity-building workshops on low-cost underwater robotics
• Explored innovative STEAM teaching methodologies and practical applications for marine science and technology education
• Shared knowledge and experiences regarding inclusive education principles and culturally responsive teaching, ensuring the DIVE outputs meet the needs of diverse student populations

Looking Ahead: Transforming STEM Education

Over the next three years, DIVE will:

• Develop an inclusive hands-on curriculum in the fields related to underwater robotics and marine science
• Train 260 teachers in innovative, interdisciplinary STEAM approaches
• Reach 4,000 students in multicultural classrooms
• Build a sustainable network of educators across Spain (Catalonia and Canary Islands), Greece, and Portugal to foster collaboration and knowledge exchange

The Girona kick-off provided a dynamic space for partners to connect, share expertise, and lay the groundwork for the project’s objectives.

Stay tuned as we dive into this exciting journey, shaping the future of STEAM education across Europe!

The post DIVE into Learning Kicks Off with First Transnational Meeting in Girona appeared first on Vicorob.

During the conference, Vicorob presented the poster “Volumetric-Guided Brain Image Synthesis with Diffusion Models”, showcasing a novel approach to generating and transforming brain MRI images using artificial intelligence.

The work introduced a method that can generate brain MRI images and transform existing ones. It allows adjusting specific brain regions to explore “what if” scenarios, such as simulating changes related to aging or neurological conditions, and can predict how brain structure may evolve over time. These capabilities provide a flexible tool for studying brain changes and supporting research in neuroscience and medical imaging.

The conference program included keynote talks by leading researchers such as Ellen Van Obberghen-Schilling, who presented advances in AI-driven spatial imaging for personalized tumor treatments; Tom Vercauteren, who explored AI-powered hyperspectral imaging for real-time surgical support; and Katrien Verbert, who discussed interactive and conversational AI to improve the understanding of AI-based decisions.

Vicorob’s participation in IABM 2026 reinforces its commitment to advancing artificial intelligence technologies with a strong focus on real-world biomedical applications and societal impact.

The post Vicorob presents new AI-based approach for brain image simulation at IABM 2026 appeared first on Vicorob.

The workshop brought together all the research teams whose projects were selected in this call, focusing on robotics and digital technologies for sustainable agriculture. MULTIFUSE, a project with participation from Vicorob, is one of the six funded projects addressing how robotics and digital technologies can support more sustainable agricultural practices.

This first meeting provided a valuable opportunity to network with other international teams, exchange perspectives, and start shaping collaboration between European and Japanese partners.

Robotics and digital technologies for sustainable agriculture

On January 31, 2025, the EIG CONCERT-Japan initiative announced the six selected projects of its 11th joint call on digital transformations and robotics in sustainable agriculture.

MULTIFUSE: artificial intelligence for plant stress detection

The MULTIFUSE project focuses on advanced multimodal sensing and data fusion for early digital detection of plant stress symptoms.

By combining data from various optical sensors, the research trains artificial intelligence to detect plant stress, such as water deficiency or infection, at a physiological level before it becomes visible to the human eye.

Other projects in robotics and agriculture

The six selected projects address sustainable agriculture through robotics, artificial intelligence, and digital technologies:

• HydroNPK-AI focuses on AI-enhanced NPK management for optimized hydroponic cultivation. Using real-time artificial intelligence, the system manages nitrogen, phosphorus, and potassium in hydroponic setups to maximize yield while reducing chemical runoff.
• DRAGONS develops drones equipped with robotic arms to autonomously collect plant samples such as leaves or fruit from complex vineyard canopies.
• DEMETER builds adaptive pruning robots through simulation, artificial intelligence, and intuitive teleoperation, allowing human experts to remotely guide robotic systems.
• DYNA-GROW focuses on dynamic lighting management systems and sustainable energy solutions for smart crop production, integrating renewable energy with intelligent lighting.
• Supporting Digital Solutions develops a framework to support the demonstration of sustainable digital solutions for agriculture and enable collaboration between Japanese and European startups.

Social sustainability in agriculture

In both Japan and Europe, the average farmer is over 60 years old. These projects aim to make agriculture physically easier and more attractive to younger generations through robotics and digital technologies.

International collaboration in robotics and sustainable agriculture

The workshop in Bonn marked the beginning of collaboration between the selected teams, enabling them to connect, exchange perspectives, and start working together in the field of robotics and sustainable agriculture.

Vicorob’s participation in MULTIFUSE is part of this international effort to explore how robotics and digital technologies can support more sustainable agricultural practices.

The post Vicorob participates in EIG CONCERT-Japan workshop on robotics and sustainable agriculture appeared first on Vicorob.