Scholarlink

Background <p>To explore the transformative role of artificial intelligence (AI) in the detailed analysis and characterization of grouped breast microcalcifications, highlighting its potential to improve the diagnostic accuracy. In the current retrospective analysis, artificial intelligence 2D solution was used to read mammograms showing grouped microcalcifications and interpretation included qualitative and quantitative information to assess suspicion of malignancy. All the included abnormal grouped calcification were pathologically proven.</p> Results <p>The&#xa0;performance of AI in providing abnormality scoring percentage&#xa0;varied depending on the type of morphology that is been evaluated. The coarse heterogeneous morphology had a cutoff value of 69.5%, with a balanced sensitivity 87.5% and specificity of 67.8% (<i>P</i> = 0.000), the fine pleomorphic morphology had a cutoff value of 87.5%, with an accepted sensitivity 63.83% and high specificity of 86.21% (<i>P</i> = 0.000). The fine linear descriptor had a higher malignancy rate of 52.94% and the lowest performance in both sensitivity and specificity.</p> Conclusions <p>Despite challenges in achieving balanced performance for grouped microcalcifications, the findings affirm the potential of AI-assisted analysis in enhancing diagnostic precision in detecting malignancy in coarse heterogeneous microcalcifications and identifying benign cases in amorphous morphology type.</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

How artificial intelligence is changing the way we understand grouped breast microcalcifications?

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月20日

查看全文
  • Sahar Mansour,
  • Abeer Ali El-Sharawy,
  • Salma Moheeb Abdelghaffar,
  • Ghada Mohammed Abdel-Salam,
  • Rehab ElSheikh

摘要

Background

To explore the transformative role of artificial intelligence (AI) in the detailed analysis and characterization of grouped breast microcalcifications, highlighting its potential to improve the diagnostic accuracy. In the current retrospective analysis, artificial intelligence 2D solution was used to read mammograms showing grouped microcalcifications and interpretation included qualitative and quantitative information to assess suspicion of malignancy. All the included abnormal grouped calcification were pathologically proven.

Results

The performance of AI in providing abnormality scoring percentage varied depending on the type of morphology that is been evaluated. The coarse heterogeneous morphology had a cutoff value of 69.5%, with a balanced sensitivity 87.5% and specificity of 67.8% (P = 0.000), the fine pleomorphic morphology had a cutoff value of 87.5%, with an accepted sensitivity 63.83% and high specificity of 86.21% (P = 0.000). The fine linear descriptor had a higher malignancy rate of 52.94% and the lowest performance in both sensitivity and specificity.

Conclusions

Despite challenges in achieving balanced performance for grouped microcalcifications, the findings affirm the potential of AI-assisted analysis in enhancing diagnostic precision in detecting malignancy in coarse heterogeneous microcalcifications and identifying benign cases in amorphous morphology type.