Speaker
Description
Liquid Argon Time Projection Chambers (LArTPCs) enable precise 3D imaging of neutrino interactions at millimeter-scale resolution, making them a leading technology for accelerator-based neutrino oscillation physics. The SPINE reconstruction chain (Scalable Particle Imaging with Neural Embeddings) leverages Sparse Convolutional Neural Networks for voxel-level feature extraction and Graph Neural Networks for particle-level clustering, forming a unified end-to-end pipeline for neutrino interaction reconstruction. We present the performance of SPINE on the ICARUS detector — a 476-ton LArTPC serving as the Far Detector of the Short Baseline Neutrino (SBN) program at Fermilab — and its critical role in enabling ICARUS’s flagship $\nu_\mu$ disappearance and $\nu_e$ appearance oscillation measurements. Michel electrons from muon decay-at-rest provide a clean, well-understood benchmark for validating reconstruction performance and energy scale calibration, serving as a stepping stone toward the full oscillation analysis.
