Additive manufacturing enables novel classes of product complexity together with high levels of customization for functional components that shall meet strict quality requirements. Because of this, novel product qualification and statistical process control challenges must be faced in the industry. Traditional methods could become not applicable in this new scenario as they are based on monitoring simple geometrical or dimensional features, and they entail a training phase consisting of several copies of the same part. Because of this, the possibility of using data and signals acquired via in situ sensors during the process to support qualification procedures has been attracting a great deal of industrial interest. This study presents an in situ metrology method that combines in situ geometry reconstruction of the part via layerwise image segmentation and a quality modeling approach that allows estimating synthetic geometric patterns in terms of one-dimensional profile data. The method is specifically designed for lattice structures, one kind of complex shape enabled by additive manufacturing processes. It aims at providing a methodological framework to anticipate geometrical assessment and anomaly detection while the part is being built and even in the presence of one-of-a-kind structures. A real case study in laser powder bed fusion is presented to demonstrate the feasibility and benefits of the proposed approach.