Prototype software successfully ‘reads’ spinal CT for fracture diagnosis

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon
 - Spine

A research team has designed and validated a software algorithm that uses quantitative analysis of CT images to automatically detect and precisely localize vertebral body fractures in the thoracic and lumbar spine.

In a study report published in the January edition of Radiology, they report that their fully automated system showed 92 percent sensitivity for fracture detection and localization of the correct vertebra, with a false-positive occurrence rate of only 1.6 per patient.

Joseph Burns, MD, PhD, of the University of California at Irvine and colleagues from the NIH’s Imaging Biomarkers and Computer-Aided Detection Laboratory in Bethesda, Md., retrospectively analyzed 104 previously obtained scans.

Of these, 94 had positive findings for fractures (59 with vertebral body fractures), while 10 had no vertebral fractures and so served as control exams.

The case set involved 141 thoracic and lumbar vertebral body fractures, and these were marked and classified by a radiologist.

Burns and colleagues divided the CT dataset into training and testing (i.e., validation) subsets (37 and 67 subsets, respectively) for analysis by means of their prototype software. They found:

  • Training-set sensitivity for detection and localization of fractures within each vertebra was 0.82 (28 of 34 findings), with a false-positive rate of 2.5 findings per patient.
  • The sensitivity for fracture localization to the correct vertebra was 0.88 (23 of 26 findings), with a false-positive rate of 1.3.
  • Testing-set sensitivity for the detection and localization of fractures within each vertebra was 0.81 (87 of 107 findings), with a false-positive rate of 2.7.
  • The sensitivity for fracture localization to the correct vertebra was 0.92 (55 of 60 findings), with a false-positive rate of 1.6.

“Although not yet at the point of clinical application, our computer system automatically detects vertebral body fractures in the thoracic and lumbar spine as an initial step toward a fracture detection system that will assess both the vertebral body and posterior elements,” Burns et al. write in their report.

They add that their software algorithm detects the level of the fractured vertebra and the precise location of the fracture within the vertebra and so “may assist the radiologist in fracture classification according to orthopedic trauma surgery classification systems.”

Further, they state, the system has the potential to “decrease interobserver variability in fracture detection and help standardize fracture reporting.”