Explainable Neuro-Symbolic Spiking Neural Networks fo rHigh-Precision Intracranial Tumor Diagnosis

Abstract

Intracranial tumor diagnosis requires exceptional precision due to the complex morphology of brain structures andtheclinicalrisks associated with misclassification. While deep learning has improved the automation of MRI-based tumor analysis, itsblack-boxnature limits trust, interpretability, and clinical deployment. This paper introduces an explainable neuro-symbolic spikingneural network(NS-SNN) framework that integrates biologically inspired spike-based computation with symbolic reasoning mechanismstoachievehigh-precision and fully interpretable intracranial tumor diagnosis. The approach leverages the temporal dynamics andenergyefficiencyof spiking neural networks while embedding medical knowledge graphs and rule-based logic to provide transparent diagnosticinsights.By bridging data-driven neural representations with human-understandable symbolic rules, NS-SNNs offer enhancedexplainability,robustness, and alignment with radiological reasoning. This work demonstrates how neuro-symbolic integration improvesdecisiontraceability, supports causal interpretation, enables uncertainty analysis, and meets emerging clinical requirements for transparentAIinhealthcare. Results indicate that NS-SNNs provide competitive diagnostic accuracy while generating interpretable reasoningpathwaysthat can be reviewed and validated by clinicians, positioning them as a promising next-generation AI paradigmfor safe andprecisebraintumor diagnosis.