Surgical Phase Recognition: A Review and Evaluation of Current Approaches

Objective: In the last decades, there has been a growing interest in exploring surgical procedures with statistical models to analyze operations in different se?mantic levels. This information is necessary for developing context-aware intelligent systems, which can assist the physicians during operations, evaluate procedures after?ward or help the management team to effectively utilize the operating room. The objective is to extract reliable patterns from surgical data for the robust estimation of surgical activities performed during operations. This paper reviews the state-of-the-art methods for the recognition of highest?level surgical activities, i.e., phases, and focuses on data and learning algorithms. Methods: Three databases, IEEE Xplore, Scopus, and PubMed are searched, and additional studies are added through a manual search. After the database search, 173 studies are screened and a total of 33 studies are selected for the review. Conclusion: The de?velopment of robust surgical phase recognition algorithms requires large and diverse datasets, as the task is highly variable and complex. Much progress has been made, but publicly available data are still limited in volume and biased to endoscopic videos. Unsupervised or semi-supervised learning approaches and active learning methods are pro?posed to lessen the consequences of these challenges. Additionally, other potential data sources in the operating room and using different modalities together could be in?vestigated in the future. Significance: The present study provides a comprehensive review of surgical phase recog?nition algorithms, their generalizability, and point under?investigated areas for possible improvements