Ultrasound is a widely used imaging technique that typically produces a sequence of 2D images. However, tracked ultrasound is an increasingly popular method of converting 2D sweeps into 3D. The tracker provides data that enables subsequent reconstruction of an ultrasound volume. However, due to factors such as measurement inaccuracy, synchronization issues, and other disruptive influences, this tracking data may not accurately provide pose with exactitude. Here, we demonstrate that modifying the tracking data using anatomical knowledge can improve the compounded volume. This was achieved using a gradient descent approach to detect and correct potential errors in the tracking data. Validating against an MRI scan and comparing metrics – such as the Dice-Sørensen coefficient, Hausdorff distance, as well as volume and surface area change – demonstrates that improvements can be achieved.