Keywords:
Image registration, Computer Applications-3D, Ultrasound, Breast
Authors:
J. C. H. Chiu1, A. Choate2, J. Ecanow2, G. Spear3, S. Mondello 4, Y. Lei5, C. Segarceanu5; 1Evanston, IL/US, 2Evanston/US, 3Evanston /US, 4Messina/IT, 5Chicago/US
DOI:
10.1594/ecr2018/C-2153
Conclusion
Our findings show that the displacement of small masses,
relative to nipple and body planes correlates with the changes in probe position and orientation,
as well as changes in body position on the exam table.
Large target displacements of a few cm were associated with larger differences in the probe and body positions between exams which are probably responsible for the limited reproducibility of lesions localization with the current manual annotations.
A localization error of approximately 7 mm can be obtained by matching the body position on table and the probe position and orientation between exams; however this level of precision requires automated coordinates.
Precise automated mapping can allow for the confident re localization of small lesions,
even in the absence of similar lesion morphology and size between exams.
Precise localization appears to be not affected by breast size or lesion location.
In addition,
precise localization of the breast tissue in multiple sequential images during scanning can support the completeness of scanning evaluation for whole breast exams.
Lesion-to-nipple distance slightly varied based on the changes of probe and body positions,
which is consistent with previous studies.
Therefore,
the distance to nipple alone is not sufficient to precisely re-localize a breast lesion.
The localization error with the current manual method was more than 4 times (up to 41 mm) larger as compared to the measurements obtained with the automated system.
Our study has also demonstrated that considerable time saving (between a factor of three to eight) can be achieved by using the automated annotation system,
thereby reducing costs and improving the workflow in a busy breast imaging center.