Development and validation of SCAISS, a tool for semi-automated quantification of sacroilitis by magnetic resonance in spondyloarthritis

Abstract

To develop a semi-automated method to quantify inflammation in sacroiliac (SI) joints by measuring bone marrow edema (BME) on MRI. The SCAISS was designed as an image-processing software. Validation followed: (1) three readers evaluated SI images of 23 patients with axial SpA and various levels of BME severity with the SCAISS, and two non-automated methods, SPARCC and Berlin; (2) 20 readers evaluated 12 patients images, also with the three methods; (3) 203 readers evaluated 12 patient images with the Berlin and the SCAISS. Convergent validity, reliability and feasibility were estimated in the first two steps and reliability was confirmed with the third. The interobserver reliability (ICC and 95% CI) in the three observers’ study was similar across methods: SCAISS = 0.770 (0.580–0.889); Berlin = 0.725 (0.537–0.860); and SPARCC = 0.824 (0.671–0.916). In the 20 observers’ study, ICC was: SCAISS = 0.801 (0.653–0.927); Berlin = 0.702 (0.518–0.882); and SPARCC = 0.790 (0.623–0.923). In the 203 observers’ study, ICC were: SCAISS = 0.810 (0.675–0.930), and Berlin = 0.636 (0.458–0.843). SCAISS showed good convergent validity (r with SPARCC = 0.760). Median time (interquartile range) employed in the reading procedure was 28 (27) seconds for the SCAISS, 14 (9) for the Berlin score, and 94 (68) for the SPARCC. The SCAISS permits a valid, reliable, and fast calculation of overall BME lesion at the SI joint on MRI images not dependent on readers’ experience.