Scholarlink

Background <p>To extract the histogram features of quantitative parameters from synthetic MRI (SyMRI) and evaluate their diagnostic performance in differentiating benign and malignant BI-RADS 4 non-mass enhancement (NME) lesions.</p> Methods <p>This single-center retrospective study enrolled consecutive patients who underwent dynamic contrast-enhanced MRI (DCE-MRI), diffusion-weighted imaging (DWI), and SyMRI between October 2019 and May 2023. Conventional imaging characteristics and the time-signal intensity curve (TIC) were evaluated on DCE-MRI. Apparent diffusion coefficient (ADC) values and histogram features were independently measured by two radiologists, with inter-observer agreement evaluated. Multivariate logistic regression was employed to construct three diagnostic models: conventional model (based on DCE-MRI features and ADC values), SyMRI model(based on histogram features) and combined model(integrating festures from both). The diagnostic performance of these models were subsequently evaluated and compared using receiver operating characteristic (ROC) curve analysis.</p> Results <p>A total of 108 patients (mean age, 48 years ± 11 [standard deviation]) were finally included. Univariate analysis indicated significant differences in distribution, internal enhancement pattern, ADC-min, ADC-mean, TIC, and several histogram characteristics between benign and malignant NME groups. Multivariable logistic regression models demonstrated that the SyMRI model exhibited significantly higher sensitivity (98.6% vs. 73.9%), negative predictive value(NPV) (95.4% vs. 53.7%), positive predictive value(PPV) (83.7% vs. 80.6%), and area under the curve (AUC) (0.821, 95% CI: 0.728–0.915 vs. 0.739, 95% CI: 0.636–0.842) compared to the conventional model The combined model achieved the best performance for differentiating BI-RADS 4 NME lesions, with an AUC of 0.851 (95% CI: 0.764–0.937).</p> Conclusions <p>Incorporating histogram features of quantitative parameters from synthetic MRI into conventional MRI improved the diagnostic performence in differentiating benign and malignant BI-RADS 4 NME lesions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Histogram analysis of relaxometry from synthetic MRI for differentiation of breast BI-RADS 4 non-mass enhancement lesions

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年10月13日

查看全文
  • Qin Li,
  • Fucai Xie,
  • Xiaomei Jiang,
  • Haitong Yu,
  • Zhaolong Zheng,
  • Pu-Yeh Wu,
  • Qingliang Niu

摘要

Background

To extract the histogram features of quantitative parameters from synthetic MRI (SyMRI) and evaluate their diagnostic performance in differentiating benign and malignant BI-RADS 4 non-mass enhancement (NME) lesions.

Methods

This single-center retrospective study enrolled consecutive patients who underwent dynamic contrast-enhanced MRI (DCE-MRI), diffusion-weighted imaging (DWI), and SyMRI between October 2019 and May 2023. Conventional imaging characteristics and the time-signal intensity curve (TIC) were evaluated on DCE-MRI. Apparent diffusion coefficient (ADC) values and histogram features were independently measured by two radiologists, with inter-observer agreement evaluated. Multivariate logistic regression was employed to construct three diagnostic models: conventional model (based on DCE-MRI features and ADC values), SyMRI model(based on histogram features) and combined model(integrating festures from both). The diagnostic performance of these models were subsequently evaluated and compared using receiver operating characteristic (ROC) curve analysis.

Results

A total of 108 patients (mean age, 48 years ± 11 [standard deviation]) were finally included. Univariate analysis indicated significant differences in distribution, internal enhancement pattern, ADC-min, ADC-mean, TIC, and several histogram characteristics between benign and malignant NME groups. Multivariable logistic regression models demonstrated that the SyMRI model exhibited significantly higher sensitivity (98.6% vs. 73.9%), negative predictive value(NPV) (95.4% vs. 53.7%), positive predictive value(PPV) (83.7% vs. 80.6%), and area under the curve (AUC) (0.821, 95% CI: 0.728–0.915 vs. 0.739, 95% CI: 0.636–0.842) compared to the conventional model The combined model achieved the best performance for differentiating BI-RADS 4 NME lesions, with an AUC of 0.851 (95% CI: 0.764–0.937).

Conclusions

Incorporating histogram features of quantitative parameters from synthetic MRI into conventional MRI improved the diagnostic performence in differentiating benign and malignant BI-RADS 4 NME lesions.