- AI
- Molecular Imaging
- CT
- X-Ray
- Ultrasound
- MRI
- Facility Management
- Mammography
Study Finds High Concordance Between AI and Radiologists for Cervical Spine Fractures on CT
Researchers found a 98.3 percent concordance between attending radiology reports and AI assessments for possible cervical spine fractures on CT, according to new research presented at the 2024 ARRS Annual Meeting.
In a study involving over 2,600 patients assessed for possible cervical spine fractures with computed tomography (CT), researchers found that artificial intelligence (AI) assessment demonstrated an 84.1 percent sensitivity rate as well as a 20.1 percent rate for false positives.
For the retrospective study, presented at the 2024 American Roentgen Ray Society (ARRS) Annual Meeting in Boston, researchers assessed the use of the deep learning software Aidoc, which has an FDA-cleared indication for cervical spine fractures, in 2,662 patients (ranging between 20 to 102 years of age) who had CT studies covering the cervical spine. The study authors compared AI assessments versus final reports from attending radiologists.
Out of the 2,662 reviewed cases, there were 119 fractures, according to the study. The researchers found that Aidoc correctly diagnosed 95 of the 119 fractures and 2,525 of the 2,543 cases involving no fracture.
In this case, the use of the deep learning software Aidoc facilitated diagnosis of a non-displaced C6 facet fracture (A) that was originally missed (B) by a radiologist. (Images courtesy of the American Roentgen Ray Society (ARRS)).
For discerning the presence of cervical spine fractures, the study authors determined that Aidoc had a 84.1 percent sensitivity rate and a 99.1 percent specificity rate. They also noted a positive predictive value (PPV) of 79.8 percent, a negative predictive value (NPV) of 99.3 percent and a 98.3 percent concordance with radiologist assessment for the deep learning software.
“AI is an important tool for radiologists to increase diagnostic accuracy and efficiency,” wrote lead study author Nicholas Manasewitsch, M.D., a third-year diagnostic radiology resident at the University of Washington, and colleagues.
However, the researchers acknowledged a false positive rate of 20.1 percent (24 out of 119 cases). They noted that these cases included a beam artifact, chronic spine fractures, sites of prior hardware removal and misidentification of atherosclerotic calcifications.
Overall, though, the study authors said the use of deep learning technologies could help alleviate burgeoning imaging volume.
“Though further analysis and external validation (are) required, deep learning algorithms, such as Aidoc, are emerging tools that could potentially assist in triaging cases and worklist management by prioritizing flagged cases,” noted Manasewitsch and colleagues.
In regard to study limitations, the authors conceded a low 4.1 percent incidence of cervical spine features in the cohort, which included trauma and non-trauma patients. They also pointed out that thin-slice bone reformats, which may reduce sensitivity for subtle fractures, were not included in every study.
Reference
1. Manasewitsch N, Chorath K, Venugopal N, Buttar A, Babatunde A, Mossa-Basha M. Validation of a deep learning algorithm in detecting cervical spine fractures on CT. Presented at the 2024 American Roentgen Ray Society (ARRS) Annual Meeting in Boston, May 5-9, Boston. Available at: https://apps.arrs.org/MeetingPortal24/ . Accessed May 5, 2024.
CT Study: AI Algorithm Comparable to Radiologists in Differentiating Small Renal Masses
May 14th 2024An emerging deep learning algorithm had a lower AUC and sensitivity than urological radiologists for differentiating between small renal masses on computed tomography (CT) scans but had a 21 percent higher sensitivity rate than non-urological radiologists, according to new research.
What a New Meta-Analysis Reveals About Fractional Flow Reserve Assessment with Computed Tomography
May 13th 2024While acknowledging variable accuracy overall with CT-derived fractional flow reserve (FFR-CT) values, researchers found that the accuracy rate increased to 90 percent for FFR-CT values greater > 0.90 and < 0.49.
Current Insights and Emerging Roles for Contrast-Enhanced Mammography
May 10th 2024In a recent lecture at the 2024 ARRS Annual Meeting, Jordana Phillips, MD, discussed the role of contrast-enhanced mammography in staging breast cancer, evaluating response to neoadjuvant chemotherapy and recalls from screening.
