Characterization of vascular patterns in endometrial cancer via optical resolution photoacoustic microscopy

利用光学分辨率光声显微镜对子宫内膜癌的血管模式进行表征

阅读:1

Abstract

SIGNIFICANCE: Accurate classification of endometrial pathology is clinically challenging due to the heterogeneous and focal nature of precancerous and malignant lesions. Vascular remodeling is closely linked to tumor progression and may serve as a biomarker for malignancy. AIM: We aim to characterize a label-free optical-resolution photoacoustic microscopy (OR-PAM) approach for high-resolution imaging and quantitative characterization and separability assessment of endometrial vasculature. APPROACH: A custom-built OR-PAM system was used to image 34 fresh uterus samples with histologically confirmed diagnoses: normal, benign, endometrial intraepithelial neoplasia (EIN), and endometrial cancer (EC). Thirty-one quantitative vascular features were extracted from structural and spectral analyses of the photoacoustic data, and five statistically significant and minimally correlated features were selected for the separability assessment framework. A pairwise cosine similarity matrix based on these features was computed to construct a weighted similarity network, which was embedded into a two-dimensional (2D) space with a force-directed layout. A logistic regression boundary was applied to the 2D embedding to evaluate separability between normal/benign and EC/EIN clusters. A logistic regression classifier was developed from a cosine similarity matrix and cross-validated using a leave-one-out strategy. RESULTS: The cosine-similarity network graph placed 39 of 40 images on the expected side of the separation boundary. The logistic regression classifier yielded an area under the ROC curve (AUC) of 0.943, demonstrating strong discrimination between normal/benign and EC/EIN groups. CONCLUSIONS: OR-PAM combined with imaging feature analysis enables robust differentiation of endometrial pathologies and demonstrates potential as a noninvasive optical biopsy tool for endometrial assessment.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。