The performance of polymeric parts fabricated via material extrusion-based additive manufacturing (MEAM) is significantly influenced by the sintering neck length between filaments. Despite its importance, elucidating the underlying mechanism governing the formation of inter-filament sintering necks is challenging due to the difficulties in monitoring temperature fields during MEAM. In this work, the thermal condition in MEAM was simplified using a printer with a tubular hotbed, coupled with a high temporal-spatial resolution infrared (IR) camera to monitor temperature field variations during printing. An advanced IR image processing method, integrating multi-directional retrieval and whole-loop average filtering, was developed to accurately measure temperature differences between filaments. By correlating these measurements with micro-CT characterizations, a quantitative relationship between temperature difference and sintering neck length was established. Through single- or multi-loop defect experiments, microscopic defect detection and analysis were conducted, achieving the prediction of sintering neck length variations at a 10 μm level with an 85 % accuracy. This work is believed to provide potential value for further accurate prediction of the mechanical properties of MEAM parts.