Repeatability and accuracy of a novel, quantitative radiographic method for differentiating normal canine sacroiliac...

Canine sacroiliac (SI) joint subluxation or luxation is most commonly diagnosed based on qualitative radiographic assessments. Aims of this two-part, retrospective, diagnostic accuracy, pilot study were to develop and evaluate a novel quantitative method based on measuring the angle between a line connecting the iliac wings and parallel lines across three anatomical landmarks (cranial endplate of L7, caudal endplate of L6, cranial endplate of L6) on a single ventrodorsal radiograph. For the first part of the study, angle measurements from a single observer were compared for 20 normal canine pelvic radiographs and 20 pelvic radiographs with SI luxation or subluxation. Mean values significantly differed between datasets (P < 0.001). The angles for the normal pelves ranged from 0.6° to 1.5°, while abnormal angles ranged from 3.8° to 7.1°. For the second part of the study, a dataset of 25 normal and 25 abnormal canine pelvic radiographs was evaluated using the novel technique by three blinded readers with varying levels of expertise at two different time points. There was excellent reliability among the three readers with an intraclass correlation (ICC) value of 0.90 and an excellent agreement between day 0 and day 30 readings with an ICC value of 0.91. It was also determined that a cut-off angle of 2.0°, using the line parallel to the cranial endplate of L6, provided overall the best accuracy, sensitivity, and specificity to differentiate normal versus abnormal pelves. These findings may be helpful for clinical cases with equivocal diagnoses and for future development of automated diagnostic tools.