Construction and application of large-scale image database in orthopedic surgery

Abstract

Databases of medical images are valuable resources not only for clinical studies such as the analysis of disease progression or a large-scale population analysis of morphological characteristics but also for those engaged in image analysis. Databases can serve as a compendium against which newly developed algorithms can be tested and a common platform for performance comparisons with existing state-of-the-art algorithms. Several database projects that have focused on certain target modalities and diseases have been successful, including the Cancer Imaging Archive, the Alzheimer's Disease Neuroimaging Initiative, and the Osteoarthritis Initiative. Here, we introduce our efforts to construct a database of medical images and treatment records of Japanese patients who underwent hip surgery. This database currently contains computed tomography images, radiographs, and the log files of a surgical navigation system, including preoperative plans, intraoperative procedures, and postoperative outcomes (alignment). Herein, we also introduce our attempts in three applications: statistical analysis of the alignment in functional (standing) position, muscle function, and statistical analysis of surgeons' expertise from the surgical log. Open access is an important aspect for the research community, but privacy is a concern, especially for large-scale databases where per-patient consent is difficult to obtain as well as with images of patients with specific diseases wherein complete de-identification is extremely difficult. We believe our effort serves as a step toward augmentation of social acceptability of the strength of medical image databases for accelerating advanced medical treatment.