Fig. 6: Quantification of the errors in prediction.
From: Wrinkle force microscopy: a machine learning based approach to predict cell mechanics from images
a Comparison of the ground truth \({f}_{x,y}^{{{{{{{{\rm{true}}}}}}}}}\) and the predicted traction \({f}_{x,y}^{{{{{{{{\rm{predict}}}}}}}}}\). Dashed line shows a condition fpredict = ftrue. Note that we randomly reduced the number of data point 1/5 for the visibility. b Correlation coefficient R between \({f}_{x,y}^{{{{{{{{\rm{true}}}}}}}}}\) and \({f}_{x,y}^{{{{{{{{\rm{predict}}}}}}}}}\). c, d Errors of the predicted traction compared to the ground truth data: c error in the traction magnitude εf and d the traction direction εθ. Note that bars show average errors ± standard deviations among the sample number n = 15, and gray dots indicate the original data. The description m.s. denotes the microscope images.
