During reactor operation and subsequent handling phases, zirconium-based cladding is exposed to high temperature, pressure, and irradiation. Concurrent hydrogen uptake through coolant-related reactions can lead to hydride precipitation, which degrades ductility; hydride reorientation from circumferential to radial morphologies further accelerates embrittlement. Previous studies have investigated hydride-induced ductility changes in zirconium alloys using hydrogen charging/heat-treatment procedures and mechanical tests such as axial tension. To quantify microstructure-sensitive behavior, representative volume element (RVE) finite element models have been widely adopted, where hydrides are typically idealized as fixed-size fiber-shaped inclusions with random distributions, and the approach has been extended to image-based RVEs. In this study, an image-based RVE framework coupled with a physics-informed neural network (PINN) was established to predict the altered mechanical response of hydride-containing zirconium alloys. Cross-sectional micrographs reported in the literature were processed using Python and OpenCV to segment hydride regions and quantify phase fractions. The extracted statistics were then used to generate three-dimensional RVEs with Monte Carlo-based hydride distributions. The PINN model implemented in PyTorch predicted displacement fields, where the loss function incorporated equilibrium constraints, constitutive relations, and periodic boundary conditions. The predicted effective properties were finally validated through comparison with available experimental results.