Abstract: Objective. Boron neutron capture therapy (BNCT) requires accurate knowledge of the boron distribution during treatment to enable reliable dosimetry and treatment verification. Compton imaging of the 478 keV prompt gamma rays emitted following neutron capture by 10B has been proposed as a promising technique for real-time boron monitoring. This work aims to experimentally evaluate the feasibility of Compton imaging under clinically relevant boron concentrations and realistic neutron-induced background conditions. Approach. A dedicated experimental campaign was performed at the Institut Laue-Langevin (ILL, Grenoble, France) using the i-TED Compton camera array. Three experimental configurations with progressively increasing neutron-induced background were investigated, including a water phantom containing 65 ppm 10B, representative of typical tumor concentrations during BNCT. Experimental measurements were complemented by detailed Geant4 Monte Carlo simulations to interpret detector performance, identify current limitations, and assess potential detector improvements. Main results. The experiments demonstrate, for the first time, Compton imaging of 10B at a clinically relevant concentration of 65 ppm under neutron irradiation. The reconstructed 478 keV gamma-ray emission was correctly localized with a spatial resolution of 22 mm (FWHM) and a signal-to-background ratio of 6.4. The study also identifies the principal limitations affecting detector performance, namely high count-rate effects and contamination from back-scattered Compton events. Dedicated analyses and simulations show that optimized event filtering substantially mitigates image degradation, while future detector developments based on pixelated scintillator arrays and improved timing capabilities are expected to significantly enhance performance under BNCT operating conditions. Significance. These results constitute the first experimental validation of Compton imaging for boron monitoring at clinically relevant concentrations and demonstrate the potential of the technique for real-time BNCT dosimetry. The identified detector improvements provide a technically grounded pathway towards clinical implementation of gamma-ray imaging-based dose monitoring in BNCT.