Scholarlink

Background <p>Soft tissue sarcomas (STS) are malignant, locally invasive neoplasms arising from mesenchymal cells. Optical coherence tomography (OCT) has been shown to identify tissue types and can be used for margin assessment during STS removal surgeries. Due to the large number of images produced, automatic algorithms are worthwhile to screen OCT images effectively.</p> Methods <p>In this paper, we report a method based on deep learning for fast and accurate intra-operative assessment of OCT images with polarization-sensitive (PS-OCT) enhancement. Based on the predictive model, we designed a method to localize cancerous areas based on a cancer probability diagnosis curve.</p> Results <p>We tested our proposed method on a canine STS surgical margin image dataset, and the result showed 0.989 Area Under the Receiver Operating Curve (AUROC) and 91% accuracy in detecting positive margins. Our case studies for diagnostic curves also showed the ability of our model to localize cancerous tissues.</p> Conclusion <p>Our model showed promise&#xa0; in detecting and localizing cancerous tissues in OCT images. The model demonstrated good performance which&#xa0;will help to&#xa0;build a more accurate and reliable intra-operative margin assessment system for STS.</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

Deep learning-assisted surgical margin assessment for canine soft tissue sarcoma based on polarization-sensitive optical coherence tomography

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月28日

查看全文
  • Yuanlong Wang,
  • Laura E. Selmic,
  • Ping Zhang

摘要

Background

Soft tissue sarcomas (STS) are malignant, locally invasive neoplasms arising from mesenchymal cells. Optical coherence tomography (OCT) has been shown to identify tissue types and can be used for margin assessment during STS removal surgeries. Due to the large number of images produced, automatic algorithms are worthwhile to screen OCT images effectively.

Methods

In this paper, we report a method based on deep learning for fast and accurate intra-operative assessment of OCT images with polarization-sensitive (PS-OCT) enhancement. Based on the predictive model, we designed a method to localize cancerous areas based on a cancer probability diagnosis curve.

Results

We tested our proposed method on a canine STS surgical margin image dataset, and the result showed 0.989 Area Under the Receiver Operating Curve (AUROC) and 91% accuracy in detecting positive margins. Our case studies for diagnostic curves also showed the ability of our model to localize cancerous tissues.

Conclusion

Our model showed promise  in detecting and localizing cancerous tissues in OCT images. The model demonstrated good performance which will help to build a more accurate and reliable intra-operative margin assessment system for STS.