Scholarlink

Purpose <p>The diagnosis of light-chain cardiac amyloidosis (AL-CA) is difficult and often requires invasive assessment by tissue biopsy. The purpose of the study was to evaluate the diagnostic performance of a machine learning approach applied to delayed [<sup>18</sup>F]florbetaben positron emission tomography (PET) uptake in identifying patients with AL-CA.</p> Methods <p>32 patients with biopsy-proven diagnosis of AL-CA (age 67 ± 10 years, 9 women) and 45 control subjects (age 74 ± 11 years, 7 women), referred with the initial clinical suspicion and later diagnosed with non-AL-CA pathology, underwent a cardiac PET/computed tomography scan. Cardiac [<sup>18</sup>F]florbetaben PET uptake was assessed using static acquisition 110&#xa0;min after radiotracer injection.&#xa0;Unsupervised k-means and Fuzzy C-means (FCM) algorithms were applied as machine learning methods and their results were compared with statistical ROC curve analysis.</p> Results <p>Radiotracer uptake showed higher SUV values in patients with AL-CA than in control subjects (p&lt;0.001). Machine Learning, specifically unsupervised FCM algorithm, proved to be an acceptable methodology for classification, with sensitivity of 0.84 and specificity of 0.88 for SUVmean, and sensitivity of 0.88 and specificity of 0.86 for SUVmax. These values were similar to the ones obtained by statistical analysis (sensitivity 0.84, specificity 0.88 for SUVmean and sensitivity 0.88, specificity 0.91 for SUVmax).</p> Conclusion <p>Machine learning analysis of delayed [<sup>18</sup>F]florbetaben cardiac uptake could help to discriminate AL-CA from mimicking conditions and could represent a noninvasive tool for the diagnosis of AL-CA, promising to avoid tissue biopsy for a certain diagnosis.</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

Diagnostic performance of a machine learning approach applied to delayed [18F]florbetaben positron emission tomography in patients with suspected light-chain cardiac amyloidosis

  • 期刊论文

  • 发表日期：

    2026年1月23日

查看全文
  • Assuero Giorgetti,
  • Maria Filomena Santarelli,
  • Dario Genovesi,
  • Giuseppe Vergaro,
  • Alberto Aimo,
  • Filippo Bargagna,
  • Annette Kusch,
  • Brunella Favilli,
  • Gabriele Buda,
  • Angela Pucci,
  • Michele Emdin

摘要

Purpose

The diagnosis of light-chain cardiac amyloidosis (AL-CA) is difficult and often requires invasive assessment by tissue biopsy. The purpose of the study was to evaluate the diagnostic performance of a machine learning approach applied to delayed [18F]florbetaben positron emission tomography (PET) uptake in identifying patients with AL-CA.

Methods

32 patients with biopsy-proven diagnosis of AL-CA (age 67 ± 10 years, 9 women) and 45 control subjects (age 74 ± 11 years, 7 women), referred with the initial clinical suspicion and later diagnosed with non-AL-CA pathology, underwent a cardiac PET/computed tomography scan. Cardiac [18F]florbetaben PET uptake was assessed using static acquisition 110 min after radiotracer injection. Unsupervised k-means and Fuzzy C-means (FCM) algorithms were applied as machine learning methods and their results were compared with statistical ROC curve analysis.

Results

Radiotracer uptake showed higher SUV values in patients with AL-CA than in control subjects (p<0.001). Machine Learning, specifically unsupervised FCM algorithm, proved to be an acceptable methodology for classification, with sensitivity of 0.84 and specificity of 0.88 for SUVmean, and sensitivity of 0.88 and specificity of 0.86 for SUVmax. These values were similar to the ones obtained by statistical analysis (sensitivity 0.84, specificity 0.88 for SUVmean and sensitivity 0.88, specificity 0.91 for SUVmax).

Conclusion

Machine learning analysis of delayed [18F]florbetaben cardiac uptake could help to discriminate AL-CA from mimicking conditions and could represent a noninvasive tool for the diagnosis of AL-CA, promising to avoid tissue biopsy for a certain diagnosis.