Understanding malignant brain tumors means understanding the ecosystems they form within the brain. These tumors are not isolated masses of cells but highly dynamic communities that interact with neurons, immune cells, and the surrounding microenvironment. To explore this complexity, we apply spatially resolved multi-omics, a new generation of technologies that combine molecular precision with spatial context. Using advanced spatial transcriptomics approaches such as Visium, Visium HD, MERFISH, and Xenium, we map the full transcriptional landscape of brain tumors directly in their native tissue. Our integrative research frameworks extend beyond gene expression. With ElectroGenomics, we link spatial gene activity to electrophysiological signals to uncover how tumor and neural networks interact. Through SPTCR-seq, we spatially profile immune cell receptors to trace the organization of the local immune landscape. Graph-based deep learning enables us to extract meaningful signals from this complexity and to generate predictive models of tumor behavior. Through this integrative, spatially informed approach, we aim to transform the understanding of brain tumors from static entities into living, evolving ecosystems.

Brain malignancies have the ability to integrate into neuronal circuits of the central nervous system. Through direct synaptic contact and paracrine communication, tumour cells interact with neurons in processes that mimic normal brain development. The diversity of interactions between malignant cells and neurons is remarkable and are widely adopted across primary brain tumors as well as brain metastasis. The degree in which gliomas integrate into the neuronal circuits of the normal brain affects patient survival, highlighting the relevance of these processes for tumor progression. Due to the complex nature of these integrated malignant systems, detailed characterization requires innovative approaches. In the MILO lab we utilize an integrated platform that allows parallel spatial electrophysiological and transcriptional profiling on one tissue termed “elecrtogenomics” that allows us to link features of functional connectivity with microenvironmental architecture and help elucidate the roles of all microenvironmental components in the successful integration of malignant cells in neuronal circuits.

Our team focuses on leveraging and enhancing the body’s natural defense mechanisms to develop innovative strategies for treating brain cancers. Utilizing cutting-edge transcriptomics and computational techniques, we dissect the tumor microenvironment to uncover how brain tumors interacts with the immune system. By understanding these complex interactions, we aim to identify new therapeutic targets and improve existing treatments. Our investigations include analyzing the interactions between brain tumors and the immune system to understand how tumor cells evade attacks by the immune cells and identify potential targets for therapeutic intervention.

We employ advanced artificial intelligence (AI) and machine learning (ML) methodologies to create in silico models that optimize and redesign personalized treatment strategies. These technologies allow us to predict the most effective treatment strategies for each patient. Here we focus on pioneering approaches such as Chimeric Antigen Receptor (CAR) T cell therapy and other cellular therapies. By modifying and enhancing CAR-T contructs and patients' T cells, we aim to achieve a sustained anti-tumor response against brain tumors, offering new hope for patients with these challenging brain cancers. 

Our primary goal is translational research—bridging the gap between the clinic and the laboratory, and vice versa—to translate scientific discoveries into real-world improvements in patient outcomes. To achieve this, we employ two cutting-edge technologies: Stimulated Raman Histology (SRH) and intraoperative sequencing.

SRH is a novel imaging technique used directly in the operating room. Within minutes, it reconstructs the tissue architecture by converting chemical signals—originating from the vibrational frequencies of molecules in the tissue—into high-resolution images interpretable by both surgeons and pathologists. We leverage deep learning models to classify cells based on their morphology and correlate these features with transcriptional and epigenetic profiles. This enables comprehensive characterization of the tumor microenvironment and allows us to predict tumor composition and epigenetic subtype from the image itself—information that can support real-time decision-making during surgery.

In parallel, we perform intraoperative sequencing. During the surgical procedure, we collect a biopsy and generate a rapid molecular profile of the tumor within an hour. This includes identifying chromosomal alterations to confirm the diagnosis, as well as assessing the methylation profile to inform surgical strategy. Together, these approaches aim to provide surgeons with timely, actionable data that can optimize treatment decisions and improve patient outcomes.