bioRxiv Science⌕ Search

Biology subjects

Kirschke, J. S.

Publications and source records attributed to Kirschke, J. S..

3 recordsLinked to original sources

MedMCP: An Open Agentic Ecosystem for Medical Imaging Workflows

Deep learning has driven rapid progress in medical image analysis, with methods for tasks such as organ segmentation, anomaly detection, and radiology report generation now matching or surpassing expert performance on several benchmarks. While many methods are released as open source, their routine adoption by clinicians and medical researchers remains limited. The bottleneck is not a lack of capable tools, but the absence of infrastructure to make them accessible to non-technical users. To break this barrier, we present MedMCP, an agentic framework that solves this last-mile problem along two axes: (1) a local agentic workspace that lets users apply state-of-the-art image analysis methods through natural language, and (2) a community-driven tool ecosystem that lets developers deploy and distribute capabilities as validated, independently testable servers. MedMCP enforces a strict orchestration-execution boundary in which a local language model translates user intent into workflow plans, while all image analysis is delegated to domain-specific tools. The framework runs entirely on-premise, every side-effecting operation passes through a human-in-the-loop permission gate, and a full provenance trail links each output to the tools, parameters, and inputs that produced it. An inspectable workspace lets users verify intermediate results rather than blindly trusting the agent, and completed sessions distill into deterministic, shareable workflows that re-execute without the agent, turning one-off interactions into reproducible analyses. This paper introduces the project's general idea and describes the design principles that guide its development. Beyond a single toolkit, we envision MedMCP as a community-driven platform where each contributed tool becomes easily accessible, compounding individual efforts into shared capability.

bioinformatics↗

Multimodally trackable and clinically translatable platform for modelling human demyelinating brain diseases by temporally dispersed chemically induced lesions in thepig brain

BackgroundDespite advances in therapy, inflammatory demyelinating diseases of the central nervous system, such as multiple sclerosis, remain important causes of morbidity among young adults. Translation of remyelinating paradigms from current murine models is encumbered by the small size and low white matter content of the brains, limiting the spatial resolution of diagnostic imaging. Large animal models might be more suited for this purpose but pose significant technological, ethical and logistical challenges. MethodWe induced reversible and targeted cerebral demyelinating lesions by controlled injection of lysophosphatidylcholine in the minipig brain. One strength of the approach is the serial induction, allowing parallel imaging of successive stages of de-/remyelination. FindingsWe demonstrate controlled, clinically unapparent, reversible and multimodally trackable brain white matter demyelination in a large animal model. Lesions were amenable to follow-up using the same clinical imaging modalities (3T magnetic resonance imaging, 11C-PIB positron emission tomography) and standard histopathology protocols as for human diagnostics, as well as electron microscopy to compare against biopsy data from two patients with cerebral demyelination. InterpretationBy employing human diagnostic tools and validating the model against data from related human diseases, our platform overcomes one important translational barrier of current animal brain demyelination models while having the potential for developing diagnostic procedures and imaging biomarkers. Remyelination and axon preservation dynamics diverge from classical rodent models. FundingThis work was supported by the DFG under Germanys Excellence Strategy within the framework of the Munich Cluster for Systems Neurology (EXC 2145 SyNergy, ID 390857198) and TRR 274/1 2020, 408885537 (projects B03 and Z01). Research in contextO_ST_ABSEvidence before this studyC_ST_ABSInflammatory demyelinating diseases of the central nervous system (CNS), targeting primarily the white matter (WM) of the brain and spinal cord, such as multiple sclerosis (MS), still represent some of the most important non-traumatic causes of disability in young adults. Current animal models based on murine species, for example, experimental autoimmune encephalomyelitis, have been demonstrated to reliably depict pathophysiological facets of human disease. However, they are nevertheless encumbered by the low WM content and the small size of murine brains, which still pose a translational barrier to diagnostic imaging tools used in a clinical context in human patients. Minipigs are increasingly being used to model human neurological diseases, as yet primarily in the context of neurodegenerative disorders. Added value of this studyHere, we establish a platform for Minipig Stereotactic White-matter Injection using Navigation by Electromagnetism (MiniSWINE) and validate such a tool in a clinical multimodal imaging and microscopy setting against biopsy and imaging data from human demyelinating disorders across different disease stages, as well as against existing and potentially emerging human diagnostic imaging. Moreover, in order to overcome the neuroanatomical challenges of stereotactic injection in the pig brain, we designed a new electromagnetic-guided tracking system whose key advantage is the direct measurement of the injection cannula tip position in situ. Another strength of our study lies in its setup, characterized by the serial induction of successive stages of de- and remyelination, allowing for multimodal assessment via imaging and histopathology or electron microscopy of multiple stages in parallel. The remyelination dynamics inferred in this context diverge from the classical rodent studies, by exhibiting incomplete remyelination at the subacute stage, persistent astroglial and microglial activation as well as a minor degree of secondary axonal degeneration. Thus, they more closely resemble human inflammatory demyelinating brain plaques. Implications of all the available evidenceWe believe that MiniSWINE links evidence from well-established demyelination-induction methods from rodent models of CNS demyelinating disorders, as well as from human imaging and biopsy data, while at the same time providing a novel platform for the potential development of diagnostic procedures, discovery of imaging biomarkers and testing of remyelinating agents in diseases such as MS. Thus, it can have particular relevance to human health in the context of future translational animal model-based research in inflammatory demyelinating disorders of the CNS. Additionally, our electromagnetic-guided injection technique may enhance stereotactic substance delivery in human neurosurgery.

neuroscience↗

Pre-operarative Multivariate Connectome Analysis for Glioma Patients

Recent advances in neuroimaging have allowed the use of network analysis to study the brain in a system-based approach. A system-based analysis of gliomas can shed light on mechanisms underlying neuronal connectivity and plasticity and the recovery process, and it could support surgical decision-making. Surgery has been shifting from image-guided surgery to a functional mapping-guided resection where several structural and functional modalities are used. However, reliable identification of eloquent areas during planning of surgical resection is still a challenge. Pre-operative language mapping performed by navigated transcranial magnetic stimulation (nTMS) is of great value as it elucidates functional cortical organization, which might be different from patient to patient due to the heterogeneity of the lesions and individual plasticity. In this paper we propose the construction of an \"effective\" speech network used for surgical decision making. This is achieved by mapping functionally relevant areas identified by nTMS on tractography-based connectomics. Subsequently we compute graph metrics on the identified speech networks of patients who show preoperative aphasia aiming to identify relevant differences between graph metrics in patients with and without preoperative aphasia. Lastly, the validity of the speech networks is examined by checking the involved graph communities.

neuroscience↗