Hand-drawn Symbol Recognition of Surgical Flowsheet Graphs with Deep\n Image Segmentation

Abstract

Perioperative data are essential to investigating the causes of adverse\nsurgical outcomes. In some low to middle income countries, these data are\ncomputationally inaccessible due to a lack of digitization of surgical\nflowsheets. In this paper, we present a deep image segmentation approach using\na U-Net architecture that can detect hand-drawn symbols on a flowsheet graph.\nThe segmentation mask outputs are post-processed with techniques unique to each\nsymbol to convert into numeric values. The U-Net method can detect, at the\nappropriate time intervals, the symbols for heart rate and blood pressure with\nover 99 percent accuracy. Over 95 percent of the predictions fall within an\nabsolute error of five when compared to the actual value. The deep learning\nmodel outperformed template matching even with a small size of annotated images\navailable for the training set.\n

Ähnliche Arbeiten