gms | German Medical Science

Objectives: Correct image analysis plays a central role in human science but comes along with many challenges like a lack of anonymization, bias, usability, especially in terms of analysis by multiple independent analyzers, sharing and merging of results etc. Tyche is a light-weight, browser based program that aims at addressing the abovementioned challenges by displaying images anonymized, in a random order and facilitating the analysis by multiple analyzers. The goal of this study was to validate Tyche for the use of orthopedic images by comparing it with a validated standard program for orthopedic planning.

Methods: In order to compare the two methods, standard orthopedic modalities like length-, angle- and area measurements as well as a score, were chosen. 103 anterior-posterior hip radiographs were randomly selected from out database and analyzed separately by 5 experienced orthopedic surgeons using Tyche and a validated software, mediCAD®, as a reference. Pearson's correlation coefficients between Tyche and mediCAD® (mediCAD Hectec GmbH, Altdorf, Germany) for each modality were calculated. The methods were also compared using Bland-Altman-Plots for length-, angle- and area measurements. A short video explaining Tyche can be found at www.tyche.expert/explain.

Results and conclusion: All 103 radiographs were analyzed 5 times using Tyche and 5 times using mediCAD®. Significant correlations ranging from r=0.77 (angle-measurements) to r=0.99 (area-measurements) were found between the two programs for all measurements. Bland-Altman-Plots confirmed non-inferiority of either method. Images were analyzed significantly faster using Tyche. No obvious transmission errors or outliers were found in results using Tyche in contrast to those from mediCAD®.

Although there are many tools available, none of them addresses all challenges in image analysis like anonymization, subjectivity, bias and usability. Here we tested a new browser based program, Tyche, by comparing results with a validated, standard tool. In summary results correlated significantly between the two methods. However, using Tyche results were acquired faster, under full anonymization and with less transmission errors. We conclude that Tyche is a validated and well fit tool for orthopedic image analysis. For the first time, a validated browser based tool enables independent analyzers worldwide to conduct efficient multi-center randomized (orthopedic) imaging studies and score validations.