Disturbances in the arthroscopic view defined with video analysis

Abstract

Optimization of arthroscopic view with technological innovation. To formulate objective measures for arthroscopic image quality. This is achieved by defining uniquely interpretable descriptions of disturbances of the arthroscopic image. Disturbances in the arthroscopic view were defined focusing on the size of the disturbance area, and the duration of a disturbance. They were categorized as Bleeding, Turbidity, Air Bubbles, Loose Fibrous Tissue, Attached Fibrous Tissue, Tissue too Close, and Instrument too Close. Ten arthroscopic knee procedures were recorded on digital film with a frame rate of 25 images per second (almost 5 h in total). Using time-action analysis, two investigators individually scored the presence and type of each disturbance in all 5 h of the ten arthroscopies. The kappa statistic was used to assess agreement between the investigators. Additionally, five other observers, who had no prior training, scored 1 min of each of the ten arthroscopies (total of 10 min) to evaluate the definitions for unique interpretability. The adjusted kappa statistic (kappa is corrected for high prevalence) shows good tester agreement for all disturbances (range 0.71-0.96), also for the observers who had no prior training. Only Turbidity shows a moderate agreement (0.59 and 0.44). This is due to the difficulty to assess the exact start and end time of Turbidity. We succeeded in describing uniquely interpretable definitions for seven disturbances. This allows objective and quantitative determination of arthroscopic image quality, which is required to improve future arthroscopic equipment and training facilities.