We present InBox, an in-situ multiple-selection and multiple-view interface for interactive exploration of dense tube-based diffusion tensor magnetic resonance imaging (DTI) visualization. Optimal experience for such exploration demands concentration on the tract of interest (TOI). InBox facilitates such concentration by designing interaction within the three-dimensional (3D) model context. It leverages the use of screen space for splitting actions. The primary interface is for box-based selection featured by DTI model sculpturing. Multiple boxes can function together using two working modes (select and removal) and two associative logics (AND and OR) to support a rich set of operations. Results from a pilot study showed that our design of InBox was on the right track and users were inclined to use one-dimensional (1D) input. A high efficacy of selection was also observed specially within non-DTI participants.