Scholarlink

<p>This study evaluated the diagnostic performance of a novel vessel tracing software that quantifies retinal vessel tortuosity in premature infants for assessing retinopathy of prematurity (ROP). Seventy-three retinal images of premature infants were subjectively classified as having “Normal” or “Abnormal” tortuosity based on arterial tortuosity of the posterior retina by seven clinicians. The software then semi-automatically traced the same arteries within circular regions of interest centered at the optic disc, with radii ranging from 2 to 20x the disc diameters. Vessel tortuosity was defined as the sum of external angles at vessel bend points. Tortuosity was significantly greater in Abnormal images within regions exceeding 4-disc diameters (<i>p</i> &lt; 0.05). Diagnostic accuracy improved with larger regions: the area under the curve increased from 0.56 at 2 times to 0.90 at 8 times and 1.00 at 14 times the diameter. At 14 times, corresponding approximately to Zone I, the software achieved 100% sensitivity and specificity for detecting abnormal tortuosity. The software accurately quantified vessel tortuosity especially within Zone I of the retina, which may reduce inter-observer variability and aid clinicians, including non-specialists, in the evaluation of ROP eyes.</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

A novel method of quantifying retinal vascular tortuosity in retinopathy of prematurity

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月22日

查看全文
  • Han Peng Zhou,
  • Taku Toyama,
  • Gen Mihara,
  • Kosuke Nakajima,
  • Ryosuke Fujino,
  • Masako Nagahara,
  • Kentaro Hayashi,
  • Koichiro Sugimoto,
  • Takashi Ueta

摘要

This study evaluated the diagnostic performance of a novel vessel tracing software that quantifies retinal vessel tortuosity in premature infants for assessing retinopathy of prematurity (ROP). Seventy-three retinal images of premature infants were subjectively classified as having “Normal” or “Abnormal” tortuosity based on arterial tortuosity of the posterior retina by seven clinicians. The software then semi-automatically traced the same arteries within circular regions of interest centered at the optic disc, with radii ranging from 2 to 20x the disc diameters. Vessel tortuosity was defined as the sum of external angles at vessel bend points. Tortuosity was significantly greater in Abnormal images within regions exceeding 4-disc diameters (p < 0.05). Diagnostic accuracy improved with larger regions: the area under the curve increased from 0.56 at 2 times to 0.90 at 8 times and 1.00 at 14 times the diameter. At 14 times, corresponding approximately to Zone I, the software achieved 100% sensitivity and specificity for detecting abnormal tortuosity. The software accurately quantified vessel tortuosity especially within Zone I of the retina, which may reduce inter-observer variability and aid clinicians, including non-specialists, in the evaluation of ROP eyes.