Image Distortion in Biplanar Slot Scanning: Patient-specific Factors.

BACKGROUND

Error within imaging measurements can be due to processing, magnification, measurement performance, or patient-specific factors. Previous length measurement studies based on radiographs have shown good intraclass correlation coefficients (ICCs) on single images; but have not assessed interimage distortion. In our study, "image distortion in biplanar slot scanning: technology-specific factors" we determined that there is minimal image distortion due to the image acquisition when using biplanar slot scanning. In this study, we aim to determine the role of patient-specific factors in image distortion, specifically evaluating interimage distortion.

METHODS

Digital radiographs and biplanar slot scanner images were reviewed in 43 magnetically controlled growing rod (MCGR) patients. Fifty-five postoperative anteroposterior digital radiographs, 184 follow-up biplanar slot-scanner scanner posteroanterior and 76 biplanar slot-scanner scanner laterals were measured by 2 residents and 1 attending. The manufacturer reported average actuator diameter of 9.02 mm was used as our reference width.

RESULTS

Overall, within image interobserver ICC were moderate to excellent (0.635 to 0.983), but the interimage ICCs were poor (0.332). Digital radiographs consistently overestimated the MCGR actuator width (mean=9.655) and biplanar slot-scanner scanner images underestimated it (mean=8.935). The measurement range was large with biplanar slot-scanner scanner posteroanterior (up to 15%) and lateral (22%) measurements and with digital radiographs (39%). Patients with abnormal muscle tone had higher degrees of measurement variability.

CONCLUSIONS

We found that neither biplanar slot scanning nor digital radiography was precise or accurate. Digital radiographs consistently overestimated MCGR actuator width and biplanar slot scanning underestimated it. The poor ICC's within and between image subtypes and large standard error of measurement reflected a magnitude of distortion that needs to be accounted for when using length measurements clinically. Unlike the clinically insignificant error that we noted in our previous study "image distortion in biplanar slot scanning: technology-specific factors" (0.5% to 1.5% of the measurement), the error noted in this study (0.2% to 38.5% of the measurement) has the potential to be clinically significant. Patients who have abnormal muscle tone had larger measurement errors, likely stemming from motion during the slot scanning process.

LEVEL OF EVIDENCE

Level III.