Innovation in Imaging Research and Tools
ACRIN® presents research results and unveils image-markup tool at RSNA annual conference.
New and emerging clinical research findings from ACRIN® were well represented among the 2,600 scientific paper and poster presentations at RSNA's 96th Scientific Assembly and Annual meeting.
The conference also gave members of the ACRIN Biomedical Imaging Informatics Committee the opportunity to showcase their newest initiative — a novel annotated image-markup tool.
ACRIN investigators delivered two presentations that featured data analysis from the ACRIN trial, "Screening Breast Ultrasound in High-Risk Women." Wendie A. Berg, M.D., Ph.D., led this multicenter study that focused on determining the role of using breast ultrasound to screen for cancer and the associated risk of an unnecessary biopsy. In secondary analyses from this trial, findings from the study's participants were identified and prospectively recorded using standard BI-RADS® terminology. These statistics also included an additional descriptor for multiple bilateral masses to note similar benign-appearing findings in both breasts.
Trial researchers also compared similar masses detected with mammography and whole-breast ultrasound with isolated, unilateral lesions. Among the 2,662 at-risk participants who underwent three rounds of screening with mammography and whole-breast ultrasound, no malignancies were identified among multiple bilateral, benign-appearing masses by either screening method. When using mammography alone, rates of malignancy were higher for isolated circumscribed or obscured masses than for similar multiple bilateral masses.
When analyzing mammographic findings and determining management, trial researchers discovered that morphology and distribution of calcifications should also be considered, as well as whether the finding is multiple, bilateral, or similar. When using ultrasound for screening, the descriptive categories between round and oval masses allowed for further refinement in the classification of benign and benign-appearing masses.
Data from ACRIN's "Digital Mammographic Imaging Screening Trial" (DMIST) were used in a secondary analysis led by Kathryn P. Lowry, B.S. In the analysis, a Markov Monte Carlo model was used to evaluate the comparative effectiveness of annual breast-cancer screening in women with BRCA-1 gene mutations when using mammography alone, digital mammography (DM) alone, or a combination of the two with MRI. Additionally, clinical surveillance without imaging was compared with six screening-strategy variations that started at four different ages (25, 30, 35, and 40) with the goal of projecting outcome for life expectancy. When mammography-induced risk of breast cancer was taken into account, the optimal strategy involved MRI beginning at age 25 and alternating DM/MRI at six-month intervals beginning at age 30.
Data from the "National Lung Screening Trial" were also presented at the annual conference by Randell L. Kruger, Ph.D., which focused on the effectiveness of radiation dose associated with chest X-ray examinations. Retrospective data from 67,641 chest X-ray examinations were included in the assessment. Acquisition parameters at the 33 screening sites and participant-specific characteristics were considered in calculating the effective dose for each chest X-ray examination. The median participant effective dose (0.0344 mSv) was consistent with prior studies, providing more detailed information about the potential for increased risk of cancer associated with CT or chest X-ray.
A third presentation shared data from a central review of brain MRI scans. This research was obtained during the ACRIN 6677 clinical trial, with the aim of addressing the difficulty associated with the assessment of tumor progression in patients treated with the anti-VEGF antibody bevacizumab. Led by the trial's principal investigator A. Gregory Sorensen, M.D., the central review involved two readers and an adjudicator. They were trained using a customized presentation and tested for comprehension prior to performing a WHO-style bidimensional radiographic assessment of progression and assessment of serial 3-D volume (both on T1-weighted measures for post-contrast images), as well as 3-D volume assessments for T2-weighted images. Researchers concluded that training and testing improve agreement in central radiologic review and that the addition of T2 imaging increases detection of progression rates in patients with recurrent glioblastoma who are being treated with anti-VEGF therapy.
Another innovative development showcased in the RSNA Quantitative Imaging Reading Room featured a demonstration of the image Physician Annotation Device (iPAD)* from ACRIN in conjunction with Stanford University. iPAD, an open-source tool that links the semantic content of a radiologic image with the image of itself, enables physicians to annotate images so that descriptions are recorded into the computer in a machine-accessible way.
Currently, when radiologists annotate the same images differently, it is difficult to compare findings with different readers. Because iPAD stores data in a quantitative manner, it automatically provides a standard annotation vocabulary across the image repository and between readers. Readers are prompted to annotate lesions consistently and to use similar terms to describe the same features. This tool was developed by the Annotated and Image Markup Project, directed by Daniel L. Rubin, M.D., M.S., and supported by Mark A. Rosen, M.D., Ph.D., with programming assistance interns from the Princeton Internships in Civic Service Program.
RSNA attendees were also presented with information about ACRIN's expanding clinical trials portfolio and how they might participate in ACRIN research that will likely be featured at future meetings.
*Editor's Note: Rubin's team coined the term "iPAD" well before the introduction of the Apple product; however, they are currently considering new names for the annotation tool.
By Nancy Fredericks