Meeting Banner
Abstract #3628

Automatic Alignment for Tumor Assessment

Alexander Brost1, Neilesh Gupta1, Christoph Seeger1, 2, Aaryani Tipirneni1, Zhaoying Han1, Sjoerd B. Vos1, 3, Julian R. Maclaren1, Matus Straka1, Nancy J. Fischbein1, Roland Bammer1

1Center for Quantitative Neuroimaging, Stanford University, Stanford, CA, United States; 2Pattern Recognition Lab, Friedrich-Alexander-University Erlangen-Nuremberg, Erlangen, Germany; 3Image Sciences Institute, University Medical Center Utrecht, Utrecht, Netherlands

The assessment of brain tumor progression or regression is an important task in neuroradiology. To standardize review, baseline 3D MRI brain scans were first registered to an atlas and kept in a research PACS system (RAPID). When patients presented for follow-up studies, these were registered to the baseline data set and displayed next to each. Comparison of tumor behavior between the two scans was also more accurate and interpreted with higher confidence. Automatic registration of 3D data for image alignment on serial studies offers a faster and more accurate assessment of changes in tumor size than the standard clinical assessment.

Abstract PDF Presentation Video

Keywords

abnormality accurate advanced advances aligned alignment although always anatomy appearance artifacts assessment assist atlas audience automatic automatically available behavior best board brain cavities certified clinical clinicians closely community comparable comparative completed complicated confidence confirmed consequences considerable cube days decisions decrease decreased degree detect diagnostic displayed enable enhancement ensured evaluate evaluated example exams facilitate facilitates faster feature field flair foundation functions future guess hill house huge identical identifying improved increasing interpretation interpreted introduced lack lesion longer longitudinal make manual manually mass median medical mind next novel offer offers often orientation outcome past patient patients pattern performance physicians physics poorly positioning post preset prior process processing quality quantitative radiologist radiologists rapid realign recent recognition regarding registered registration regression rendering required resection resolution reviewed scale scenarios sciences serial series sets simple slice slices slicing spend stable studies system target task technologists technology template thalamus tool tools trajectory transversal treatment tumor tumors view widely woods