Video Context Improves Performance in Identifying Operative Planes on Static Surgical Images

BACKGROUND: Correct identification of the surgical tissue planes of dissection is paramount at the operating room, and the needed skills seem to be improved with realistic dynamic models rather than mere still images. The objective is to assess the role of adding video prequels to still images taken from operations on the precision and accuracy of tissue plane identification using a validated simulation model, considering various levels of surgeons' experience.

METHODS: A prospective observational study was conducted involving 15 surgeons distributed to three equal groups, including a consultant group [C], a senior group [S], and a junior group [J]. Subjects were asked to identify and draw ideal tissue planes in 20 images selected at suitable operative moments of identification before and after showing a 10- second videoclip preceding the still image. A validated comparative metric (using a modified Hausdorff distance [%Hdu] for object matching) was used to measure the distance between lines. A precision analysis was carried out based on the difference in %Hdu between lines drawn before and after watching the videos, and between-group comparisons were analyzed using a one-way analysis of variance (ANOVA). The analysis of accuracy was done on the difference in %Hdu between lines drawn by the subjects and the ideal lines provided by an expert panel. The impact of videos on accuracy was assessed using a repeated-measures ANOVA.

RESULTS: The C group showed the highest preciseness as compared to the S and J groups (mean Hdu 9.17±11.86 versus 12.1±15.5 and 20.0±18.32, respectively, p <0.001) and significant differences between groups were found in 14 images (70%). Considering the expert panel as a reference, the interaction between time and experience level was significant ( F (2, 597) = 4.52, p <0.001). Although the subjects of the J group were significantly less accurate than other surgeons, only this group showed significant improvements in mean %Hdu values after watching the lead-in videos ( F (1, 597) = 6.04, p = 0.014).

CONCLUSIONS: Adding video context improved the ability of junior trainees to identify tissue planes of dissection. A realistic model is recommended considering experience-based differences in precision in training programs.