Skip to main content

Table 3 RMSE (10e-4), PSNR, SSIM, FSIM Index for region of interest images in Fig. 7

From: Energy enhanced tissue texture in spectral computed tomography for lesion classification

MetricsMethodsReconstructed images (channel number)
1th2th3th4th5th
RMSESART6.537372,1689.11349.802010.6390
TV6.30685.72235.95626.01015.9182
LRTV6.41445.50965.58945.47895.3151
TDL6.76635.47435.47705.23264.9444
LRTP4.74313.85283.85174.03984.2506
PSNRSART30.62429.05427.02726.27025.401
TV30.92231.06930.72130.51830.495
LRTV30.77531.39831.27331.32231.429
TDL30.31131.45331.44931.72132.057
LRTP33.39734.50534.50733.96933.370
SSIMSART0.99940.99930.99890.99870.9984
TV0.99920.99930.99930.99930.9993
LRTV0.99920.99930.99930.99940.9994
TDL0.99920.99940.99940.99950.9995
LRTP0.99950.99970.99970.99970.9997
FSIMSART0.99810.99750.99580.99470.9920
TV0.99500.99410.99420.99500.9952
LRTV0.99440.99430.99420.99480.9952
TDL0.99630.99700.99700.99730.9975
LRTP0.99880.99920.99920.99910.9989
  1. TV Total variation, LRTV Low rand total variation, TDL Tensor dictionary learning, SART Simultaneous algebraic reconstruction technique, LRTP Low rank texture prior, RMSE Root of mean square error, PSNR Peak signal to noise ratio, SSIM Structure similarity index, FSIM Feature similarity index