Scholarlink

<p>Accurate, early identification of transthyretin cardiac amyloidosis (ATTR-CA) is challenging yet critical for effective treatment. In this work, the modeled endpoint is scan positivity on <InlineEquation ID="IEq3"> <EquationSource Format="TEX">\([^{99m}\text {Tc}] \text { Tc-PYP}\)</EquationSource> </InlineEquation> scintigraphy, defined by the semi-quantitative Perugini visual grade (Grade 2-3 versus Grade 0-1). Two multimodal deep-learning frameworks, including Late Fusion (LF) and Cross-Modal Fusion Network (CMF-Net), are proposed to combine <InlineEquation ID="IEq4"> <EquationSource Format="TEX">\([^{99m}\text {Tc}]\text { Tc-Pyrophosphate } ([^{99m}\text {Tc}] \text { Tc-PYP})\)</EquationSource> </InlineEquation> scintigraphy with clinical metadata for automated detection. On a curated cohort of 109 patients (62 positive, 47 negative), fusion models consistently outperformed image-only convolutional neural networks (CNNs): CMF-Net raised average accuracy by 6.9 percentage points and LF by 5.4. EfficientNet CMF-Net achieved peak accuracy 90.9% and F1-score 91.4%. Notable gains included a +30.8 percentage points sensitivity(recall) for ResNet-34 with CMF-Net and ResNet-50 LF sensitivity of 94.9%. These results show that integrating imaging and text/numeric clinical data yields superior, reproducible detection of <InlineEquation ID="IEq5"> <EquationSource Format="TEX">\([^{99m}\text {Tc}] \text { Tc-PYP}\)</EquationSource> </InlineEquation> scan positivity and may streamline scan interpretation.</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

Automated identification of cardiac amyloidosis using cross-modal neural networks on \([^{99m} \text {Tc}] \text { Tc}\)-Pyrophosphate SPECT imaging and clinical data

  • 期刊论文

  • 发表日期：

    2026年7月22日

查看全文
  • Fatih Batı,
  • Nilüfer Bıçakcı,
  • Musa Aydın,
  • Zeki Kuş,
  • Berna Kiraz

摘要

Accurate, early identification of transthyretin cardiac amyloidosis (ATTR-CA) is challenging yet critical for effective treatment. In this work, the modeled endpoint is scan positivity on \([^{99m}\text {Tc}] \text { Tc-PYP}\) scintigraphy, defined by the semi-quantitative Perugini visual grade (Grade 2-3 versus Grade 0-1). Two multimodal deep-learning frameworks, including Late Fusion (LF) and Cross-Modal Fusion Network (CMF-Net), are proposed to combine \([^{99m}\text {Tc}]\text { Tc-Pyrophosphate } ([^{99m}\text {Tc}] \text { Tc-PYP})\) scintigraphy with clinical metadata for automated detection. On a curated cohort of 109 patients (62 positive, 47 negative), fusion models consistently outperformed image-only convolutional neural networks (CNNs): CMF-Net raised average accuracy by 6.9 percentage points and LF by 5.4. EfficientNet CMF-Net achieved peak accuracy 90.9% and F1-score 91.4%. Notable gains included a +30.8 percentage points sensitivity(recall) for ResNet-34 with CMF-Net and ResNet-50 LF sensitivity of 94.9%. These results show that integrating imaging and text/numeric clinical data yields superior, reproducible detection of \([^{99m}\text {Tc}] \text { Tc-PYP}\) scan positivity and may streamline scan interpretation.