The sagittal alignment of the cervical spine plays a key role in the balance of the head and surgical treatment of cervical spinal disorders. Deviations from the physiological profile correlate with clinical symptoms, making an exact analysis of cervical balance parameters essential for preoperative planning and postoperative evaluation.
However, the manual measurement of the individual parameters is time-consuming and dependent on the physicians’ experience. Using a fully automated algorithm based on artificial intelligence (AI) could save time in everyday clinical practice and raise awareness of pathologies.
Fully automated, AI-based algorithm
With the aim of developing a method that would facilitate or even replace routine manual measurements by clinicians, RAYLYTIC’s AI team cooperated with clinicians from Waldkliniken Eisenberg and scientist at Leipzig University of Applied Sciences (Hochschule für Technik, Wirtschaft und Kultur/HTWK).
Pre- and postoperative lateral cervical X-rays of 129 patients undergoing anterior cervical discectomy and fusion or cervical disk arthroplasty were measured manually as well as blinded by two independent spinal surgeons. All parameters (C2-C7 lordosis, C1-C7 sagittal vertical axis (SVA), C2-C7 SVA, C7 slope) were measured twice by both human raters and compared to the automated measurement by the AI algorithm consisting of four interlinked convolutional neural network models.
Intra- and inter-rater reliability were quantified by mean errors (95% confidence interval (CI), standard deviation) and single-measure intra-class correlation coefficients (ICC) for absolute agreement.
Excellent results with automated measurements
With reference to Cicchetti (Psychological Assessment, 1994), ICC values greater than 0.75 were considered excellent. In the study, ICCs for intra- (range: 0.92–0.99) and inter-rater (range: 0.91–0.99) reliability reflected excellent agreement between human raters for all parameters pre- and postoperatively.
The automated measurement was possible in 95% of preoperative and 88% of postoperative images. ICC values for the agreement between the automated and human measurements were excellent for all parameters, ranging between 0.87 and 0.99 preoperatively, and 0.83 and 0.98 postoperatively.
Whitecloud Award at IMAST 2022
First results of the study were presented during last year’s annual meeting of the German Society for Spine Surgery (DWG). Further results will be presented during IMAST 2022, taking place 6–9 April 2022 in Miami, FL, USA.
IMAST is an event organized by the Scoliosis Research Society (SRS), an international society that is committed to research and education in the field of spinal deformities and has gained recognition as one of the world’s premier spine societies.
The Whitecloud Awards are given to both the best basic science and clinical papers presented at the IMAST meeting. Named after Dr. Thomas E. Whitecloud, co-founder of IMAST, the nominees are selected by the IMAST Committee from the submitted abstracts for the IMAST and the SRS Annual Meeting.