We have recently shown that the appearance of Axial-Shear Strain Elastograms (ASSEs) for the case of loosely-bonded, elliptical inclusions (like fibroadenomas in the breast) is unique and therefore has the potential to distinguish benign fibroadenomas from malignant tumors in the breast. The ASSEs were obtained using quasi-static axial compressions, in a like manner as in normal axial-strain elastography. However, strict axial compression is achieved most often only by computer-controlled acquisitions and not by more practical freehand acquisitions. In a freehand acquisition, the frame sequence may contain several frames that do not experience strict axial compression but may also experience rotation or shear deformations. In this paper, we demonstrate the importance of accounting for the type of deformation applied to a target tissue for the correct interpretation of the resulting ASSEs. Using freehand acquired in vivo examples, we show that such a frame experiencing rotation or shear deformations results in ASSEs that may potentially be misinterpreted. This may be far more detrimental compared to the corresponding axial elastogram frames that may only suffer from inferior image quality in terms of contrast-to-noise ratio (CNR). Further, we show that we may be able to eliminate these frames from a sequence of freehand acquired in vivo breast lesion data by implementing a special filtering scheme, thus significantly improving the reliability of the remaining ASSE frames. This work further suggests that under freehand conditions, frames have to be checked for the presence of undesirable deformations.