Zhang, P.; Frosio, T.

Accurate contrast transfer function (CTF) estimation is a critical first step in cryo-electron tomography (cryoET), enabling recovery of information from thick, heterogeneous specimens. This challenge is especially acute in in situ cryoET, where macromolecular complexes and disease-relevant structures are imaged at high tilt and low signal-to-noise ratio. Initial estimates CTF parameters remain difficult to automate and are essential for downstream processing. Here, we present Quinoa, a software package for robust tilt-series CTF estimation. Quinoa validates tilt geometry, assesses data quality, and generates initial estimates of defocus and phase shift, which are then refined in a global model to fit per-image defoci, astigmatism, phase shifts, specimen orientation (rotation, tilt, and pitch), and specimen thickness. A key feature is equiphase-binned polar power spectra, which reduce computational cost without compromising accuracy. In benchmarks against Warp, Ctfplotter, CTFMeasure, and AreTomo, Quinoa was most robust across challenging conditions, including severe astigmatism, high inclination, and variable phase shift. GPU acceleration enables real-time monitoring and high-throughput processing, supporting accurate in situ structural analysis of cells and tissues.