Tyche: a free novel web-based solution for efficient and unbiased analysis of scientific images

Background: Scientific image analysis is crucial in many research fields but it can be hampered by subjectivity and bias. Thus, image analysis should be performed anonymized by multiple observers and images displayed in a random order. Yet in radiology for example most PACS-viewers display patient metadata which could influence the analysis. Although desirable, including multiple observers can be logistically difficult as files have to be provided including spreadsheets to document the analyzed parameters. Methods To address these challenges we created a browser-based program displaying scientific images anonymized and in a random order. Observers are invited by the project master via a project-specific URL that is only temporarily valid. The project master can design an unlimited number of project-specific questions which will be displayed next to each image. As the analysis is performed online, results from all observers are summarized in real-time and the data are immediately visible to the project master. In this study, we present the process of analysis using histological and radiological images and compare Tyche with the traditional “offline” desktop bound way. Results Tyche allows users to upload anonymized images and create single choice and numeric questions specifically for these images. Via temporarily active URLs, access to the data can be provided to observers who can analyze images within any web browser using the integrated tools (zoom, length, etc.). Next to each image, the defined questions and answers are displayed and results are immediately visible. Aspects of traditional image analysis like laborious file sharing via clouds or e-mail with observers and data gathering via potentially error-prone spreadsheets become obsolete. Performing a histologic image analysis, no difference in results comparing Tyche with the offline analysis were found. Conclusion Tyche is a lightweight, web-based tool that facilitates anonymized image analysis by multiple observers and reduces subjectivity and bias.