Table 2 ROC-AUC comparison for various feature combinations. Values are reported as mean ± standard deviation. A value of 1.0 represents perfect performance. The table is divided into sections for readability.

From: Machine learning-assisted event classification in cadmium zinc telluride positron emission tomography detectors leveraging entanglement-informed angular correlations

Class

Config

\(\Delta \phi\)

\(\theta\)

\(\theta + \Delta \phi\)

xyz

\(xyz+\theta\)

\(xyz+\theta +\Delta \phi\)

E

xyz+E

0

0

0.68 ±0.008

0.78 ±0.008

0.79 ±0.004

0.88 ±0.006

0.77 ±0.014

0.79 ±0.03

0.86 ±0.006

0.87 ±0.05

1

0.81 ±0.006

0.72 ±0.007

0.73 ±0.005

0.84 ±0.006

0.75 ±0.014

0.77 ±0.04

0.90 ±0.003

0.88 ±0.05

2

0.71 ±0.002

0.77 ±0.007

0.79 ±0.005

0.82 ±0.009

0.77 ±0.014

0.78 ±0.02

0.86 ±0.009

0.79 ±0.03

3

0.67 ±0.018

0.74 ±0.01

0.76 ±0.004

0.81 ±0.01

0.74 ±0.017

0.76 ±0.04

0.85 ±0.006

0.77 ±0.03

1

4

0.91 ±0.001

0.80 ±0.0007

0.81 ±0.0005

0.86 ±0.0004

0.78 ±0.0006

0.80 ±0.05

0.95 ±0.0001

0.92 ±0.08

Class

Config

\(xyz+\Delta \phi\)

\(xyz+E+\Delta \phi\)

\(xyz+E+\theta\)

\(xyz+E+\theta +\Delta \phi\)

\(E+\theta\)

\(E+\Delta \phi\)

\(E+\theta +\Delta \phi\)

 

0

0

0.88 ±0.001

0.85 ±0.04

0.76 ±0.03

0.82 ±0.04

0.78 ±0.02

0.87 ±0.005

0.77 ±0.01

 

1

0.84 ±0.002

0.90 ±0.04

0.76 ±0.03

0.85 ±0.06

0.77 ±0.03

0.92 ±0.004

0.75 ±0.01

 

2

0.82 ±0.002

0.83 ±0.03

0.72 ±0.06

0.82 ±0.04

0.75 ±0.03

0.87 ±0.008

0.77 ±0.01

 

3

0.81 ±0.005

0.81 ±0.03

0.74 ±0.02

0.82 ±0.05

0.73 ±0.05

0.87 ±0.006

0.75 ±0.01

 

1

4

0.86 ±0.0006

0.91 ±0.05

0.79 ±0.06

0.91 ±0.08

0.81 ±0.05

0.95 ±0.0002

0.79 ±0.0006