Skip to main content
. 2024 Jul 25;14(15):1602. doi: 10.3390/diagnostics14151602

Table A4.

List of features used in the logistic regression models. These features were obtained using two sets of DEVICE measurements from paired mass images when the same mass was measured in different conditions. For the intraoperator analysis, the pair of measurements were obtained from the same operator, whereas for the interoperator analysis, they were obtained by 2 different operators. The target displacement threshold of 10 mm was used as the binary label of logistic regression models to indicate whether a displacement was large, and all other features were used as predictors.

Features Description Controllable
Patient rotation R difference R1R2 Yes
Patient rotation E difference E1E2 Yes
Patient rotation H difference H1H2 Yes
Probe axial angle difference β1β2 Yes
Probe EA difference EA1EA2^ Yes
Probe orientation difference α1α2 Yes
2D probe center difference P1P2 Yes
3D probe center difference P1P2 Yes
Probe height difference P_height1P_height2 Yes
Probe CFP difference P_CFP1P_CFP2 Yes
Probe ratio difference P_ratio1P_ratio2 Yes
Probe 2DDN difference P2DDN1P2DDN2 Yes
Probe 3DDN difference P3DDN1P3DDN2 Yes
Greater target to skin maxT_skin1, T_skin2 No
Greater target to chest maxT_chest1,T_chest2 No
Greater target 3DDN maxT3DDN1,T3DDN2 No
Greater target 2DDN maxT2DDN1,T2DDN2 No
Greater target CFP maxT_CFP1,T_CFP2 No
Greater lesion volume maxT_volume1,T_volume2 No
Target displacement T1T2

† The angle between two sets of Euler angles was calculated by determining the angular distance between corresponding quaternions.