Skip to main content

Table 2 p values for the association between each feature and IC versus SC, from logistic regression modelling, adjusted for mammographic percent destiny

From: Novel mammographic image features differentiate between interval and screen-detected breast cancer: a case-case study

 

p value, n = 1401 (1C: 393, SC: 1008)

 
 

Dense area selection method

Image feature

(1) Percentile

(2) Entropy

(3) Otsu's method

Technical feature description - all listed features are based on the identified dense area

F1

0.058

0.069

0.575

DC coefficient of two-dimensional discrete cosine transform

F2

0.636

0.006

0.233

The approximation coefficient of the multilevel discrete two-dimensional wavelet transform

F3

0.002

0.660

0.006

The max coefficient of the two-dimensional discrete Fourier transform of the local range filter

F4

0.193

0.697

0.037

Kurtosis of the intensity histogram

F5

0.388

0.334

0.053

Skewness of the intensity histogram

F6

0.074

0.026

0.015

Entropy of the intensity histogram

F7

0.282

0.031

0.969

The mean of entropy-filtered selected area

F8

0.467

0.036

0.793

Entropy of the complex imaginary part of the convolved selected area with log-Gabor filters

F9

0.624

0.045

0.969

Entropy of the complex real part of the convolved selected area with log-Gabor filters

F10

0.582

0.025

0.676

Entropy of the magnitude part of the convolved selected area with log-Gabor filters

F11

0.020

0.148

0.235

The max coefficient of the two-dimensional discrete Fourier transform

F12

0.995

0.026

0.234

Entropy of the normalized co-occurrence matrix

F14

0.600

0.751

0.326

The max coefficient of the two-dimensional discrete Fourier transform of the Hessian filter

F16

0.058

0.088

0.582

The max coefficient of the log of the magnitude part of the discrete cosine transform

F17

0.020

0.146

0.235

The max coefficient of the log of the magnitude part of the discrete Fourier transform

F18

0.111

0.078

0.443

The fourth central moment

F19

0.012

0.726

0.019

Number of separate blobs

F20

0.312

0.015

0.129

Solidity

F21

0.001

0.970

0.007

Eccentricity

F22

0.966

0.240

0.365

Euler number

F24

<0.001

0.931

0.006

Skewness of the normalized singular value decomposition/its standard deviation

F25

0.040

0.164

0.291

Singular value decomposition

F26

0.062

0.269

0.795

Number of holes within the selected area (low intensity surrounded by high intensity)

F27

0.063

0.026

0.009

Interquartile range of the intensity histogram

F33

0.225

0.317

0.322

Kurtosis of the projection along the Y axis

F34

0.065

0.136

0.201

Kurtosis of the projection along the X axis

F35

0.002

0.211

0.582

Perimeter of the selected area

F37

0.317

0.040

0.415

Mean intensity

F39

0.358

0.168

0.283

Median intensity

F40

0.145

0.012

0.001

Skewness of the intensity gradient

F41

0.172

0.032

0.946

Energy property of the co-occurrence matrix of the selected area - horizontal shift two pixels

F42

0.261

0.031

0.984

Energy property of the co-occurrence matrix of the selected area - diagonal shift eight pixels

  1. p values in bold font signifies that they are below 0.05. Only features based on the dense area of the mammogram were included
  2. 1C interval breast cancer, SC screen-detected breast cancer