Skip to main content

Table 3 Model comparisons in term of AUC, net reclassification index (NRI) and integrated discrimination improvement (IDI) in the internal ACRIN 6698 test cohort and the external test cohort

From: Early prediction of neoadjuvant therapy response in breast cancer using MRI-based neural networks: data from the ACRIN 6698 trial and a prospective Chinese cohort

Model comparisons

Delong

NRI

IDI

p

95%CI

p

95%CI

p

ACRIN 6698 test

     

 MESN-C vs. Clinical

0.007

0.329 (0.036–0.622)

0.027

0.298 (0.178–0.418)

< 0.001

 MESN vs. Pre

0.001

0.374 (0.083–0.666)

0.012

0.119 (0.038–0.200)

0.004

 MESN-C vs. Pre-C

0.001

0.205 (-0.055–0.465)

0.123

0.255 (0.120–0.390)

< 0.001

 MESN vs. MESN (T1)

0.006

0.296 (0.063–0.529)

0.013

0.117 (0.070–0.164)

< 0.001

 MESN-C vs. MESN-C (T1)

0.014

0.283 (0.044–0.522)

0.020

0.167 (0.046–0.289)

0.007

 MESN vs. MESN (T1 + DW)

0.004

0.311 (0.080–0.542)

0.008

0.076 (0.029–0.122)

0.002

 MESN-C vs. MESN-C (T1 + DW)

0.017

0.122 (-0.051–0.295)

0.167

0.161 (0.062–0.259)

0.001

 MESN vs. MESN (T1 + ADC)

0.704

0.013 (-0.210–0.238)

0.905

0.052 (0.017–0.088)

0.004

 MESN-C vs. MESN-C (T1 + ADC)

0.253

0.228 (0.014–0.443)

0.037

0.162 (0.077–0.246)

< 0.001

 MESN vs. NoESM

0.004

0.387 (0.171–0.603)

< 0.001

0.089 (0.033–0.146)

0.002

 MESN-C vs. NoESM-C

0.007

0.280 (0.048–0.513)

0.018

0.238 (0.123–0.346)

< 0.001

External test

     

 MESN-C vs. Clinical

< 0.001

0.235 (0.111–0.358)

< 0.001

0.224 (0.164–0.284)

< 0.001

 MESN vs. Pre

< 0.001

0.266 (0.110–0.421)

0.001

0.086 (0.053–0.119)

< 0.001

 MESN-C vs. Pre-C

< 0.001

0.256 (0.112–0.401)

0.001

0.194 (0.124–0.264)

< 0.001

 MESN vs. MESN (T1)

0.003

0.101 (-0.018–0.222)

0.096

0.068 (0.043–0.094)

< 0.001

 MESN-C vs. MESN-C (T1)

0.024

0.121 (0.015–0.228)

0.026

0.086 (0.032–0.140)

0.002

 MESN vs. MESN (T1 + DW)

0.007

0.051 (-0.065–0.168)

0.388

0.036 (0.012–0.061)

0.004

 MESN-C vs. MESN-C (T1 + DW)

0.030

0.100 (-0.008–0.209)

0.071

0.057 (0.008–0.106)

0.022

 MESN vs. MESN (T1 + ADC)

0.057

0.004 (-0.101–0.110)

0.939

0.056 (0.040–0.073)

< 0.001

 MESN-C vs. MESN-C (T1 + ADC)

0.212

0.017 (-0.054–0.089)

0.633

0.120 (0.084–0.157)

< 0.001

 MESN vs. NoESM

0.058

0.140 (0.035–0.245)

0.009

0.025 (0.004–0.046)

0.020

 MESN-C vs. NoESM-C

0.058

0.061 (-0.035–0.156)

0.211

0.112 (0.063–0.161)

< 0.001

  1. Note. Significant p-values (p < 0.05) are highlighted in bold. Clinical: model based on clinicopathological characteristics; Pre: model built based on pre-features; Pre-C: model built using the signature of Pre and clinicopathological characteristics; MESN: MRI-based enhanced self-attention network; MESN-C: model built using the signature of MESN and clinicopathological characteristics. NoESM: MESN model without enhanced self-attention module (ESM); NoESM-C: model built using the signature of NoESM and clinicopathological characteristics. The parentheses after MESN or MESN-C contain the MRI sequences input into the model
  2. T1 = the peak phase of T1-weighted dynamic contrast enhanced MRI; DW = diffusion weighted imaging; ADC = apparent diffusion coefficient
Back to article page

Breast Cancer Research

ISSN: 1465-542X

Contact us