Scholarlink

Radiomics analysis was used in this study to examine whether it could be possible to distinguish the carotid arteries of sickle cell disease mice (SS) from heterozygous controls (AS) using contrast-free carotid magnetic reso-nance imaging (MRA) images. A total of 66 MRA scans of one-month-old mice and 64 scans of three-month-old mice were used. Using PyRadiomics software, PyRadiomics preprocessing parameters critical for reproducibility include: version 3.0 (IBSI-compliant) 112 radiomic features were extracted from segmented carotid ar-tery images, followed by radiomic feature selection, K-fold splitting of the dataset for cross-validation, and training and validation of a predictive mod-el. At 1 month of age, four radiomic features (Original glcm Imc1; Original firstorder Minimum; Original shape MinorAxisLength; Original shape Ma-jorAxisLength) yielded an accuracy of 74% (AUROC of 0.72; 95% C.I. be-tween 0.60 and 0.84 with a p value of 0.032). At three months, a single fea-ture (Original shape MeshVolume) achieved an accuracy of 76% (AUROC of 0.67; 95% CI between 0.61 and 0.87; p-value of 0.0020). This study shows the efficacy of radiomics in distinguishing arterial features between SCD and control mice. A noninvasive and translational method may be able to assess arterial alterations in SCD for prognostic purposes in the manage-ment of this disease also in humans.

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

Radiomic Analysis of Mouse Magnetic Resonance Images and Correlation of Features Extracted to Vascular Defects in Sickle Cell Disease: Preclinical Applications

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Viviana Benfante,
  • Liana Hatoum,
  • Hannah Song Lee,
  • Edward A. Botchwey,
  • Manu O. Platt,
  • Albert Comelli

摘要

Radiomics analysis was used in this study to examine whether it could be possible to distinguish the carotid arteries of sickle cell disease mice (SS) from heterozygous controls (AS) using contrast-free carotid magnetic reso-nance imaging (MRA) images. A total of 66 MRA scans of one-month-old mice and 64 scans of three-month-old mice were used. Using PyRadiomics software, PyRadiomics preprocessing parameters critical for reproducibility include: version 3.0 (IBSI-compliant) 112 radiomic features were extracted from segmented carotid ar-tery images, followed by radiomic feature selection, K-fold splitting of the dataset for cross-validation, and training and validation of a predictive mod-el. At 1 month of age, four radiomic features (Original glcm Imc1; Original firstorder Minimum; Original shape MinorAxisLength; Original shape Ma-jorAxisLength) yielded an accuracy of 74% (AUROC of 0.72; 95% C.I. be-tween 0.60 and 0.84 with a p value of 0.032). At three months, a single fea-ture (Original shape MeshVolume) achieved an accuracy of 76% (AUROC of 0.67; 95% CI between 0.61 and 0.87; p-value of 0.0020). This study shows the efficacy of radiomics in distinguishing arterial features between SCD and control mice. A noninvasive and translational method may be able to assess arterial alterations in SCD for prognostic purposes in the manage-ment of this disease also in humans.