IEEE Trans Biomed Eng. 2019 Sep;66(9):2447-2456. doi: 10.1109/TBME.2018.2890167. Epub 2019 Jan 1.
Ultrahigh-resolution optical coherence microscopy (OCM) has recently demonstrated its potential for accurate diagnosis of human cervical diseases. One major challenge for clinical adoption, however, is the steep learning curve clinicians need to overcome to interpret OCM images. Developing an intelligent technique for computer-aided diagnosis (CADx) to accurately interpret OCM images will facilitate clinical adoption of the technology and improve patient care.
497 high-resolution three-dimensional (3-D) OCM volumes (600 cross-sectional images each) were collected from 159 ex vivo specimens of 92 female patients. OCM image features were extracted using a convolutional neural network (CNN) model, concatenated with patient information [e.g., age and human papillomavirus (HPV) results], and classified using a support vector machine classifier. Ten-fold cross-validations were utilized to test the performance of the CADx method in a five-class classification task and a binary classification task.
An 88.3 ± 4.9% classification accuracy was achieved for five fine-grained classes of cervical tissue, namely normal, ectropion, low-grade and high-grade squamous intraepithelial lesions (LSIL and HSIL), and cancer. In the binary classification task [low-risk (normal, ectropion, and LSIL) versus high-risk (HSIL and cancer)], the CADx method achieved an area-under-the-curve value of 0.959 with an 86.7 ± 11.4% sensitivity and 93.5 ± 3.8% specificity.
The proposed deep-learning-based CADx method outperformed four human experts. It was also able to identify morphological characteristics in OCM images that were consistent with histopathological interpretations.
Label-free OCM imaging, combined with deep-learning-based CADx methods, holds a great promise to be used in clinical settings for the effective screening and diagnosis of cervical diseases.
超分辨率光学相干显微镜（OCM）最近已显示出其在准确诊断人类宫颈疾病方面的潜力。然而，其在临床应用中面临的一个主要挑战是临床医生需要克服陡峭的学习曲线才能解释 OCM 图像。开发用于计算机辅助诊断（CADx）的智能技术来准确解释 OCM 图像将有助于该技术在临床上的应用，并改善患者的护理。
从 92 名女性患者的 159 个离体标本中采集了 497 个高分辨率三维（3-D）OCM 体数据集（每个体数据集中有 600 个横截面图像）。使用卷积神经网络（CNN）模型提取 OCM 图像特征，将其与患者信息（例如年龄和人乳头瘤病毒（HPV）结果）串联，并使用支持向量机分类器进行分类。使用十折交叉验证测试 CADx 方法在五类分类任务和二类分类任务中的性能。
实现了五类宫颈组织的精细分类，即正常、外翻、低级别和高级别鳞状上皮内病变（LSIL 和 HSIL）和癌症，分类准确率为 88.3 ± 4.9%。在二类分类任务[低危（正常、外翻和 LSIL）与高危（HSIL 和癌症）]中，CADx 方法的曲线下面积值为 0.959，灵敏度为 86.7 ± 11.4%，特异性为 93.5 ± 3.8%。
所提出的基于深度学习的 CADx 方法优于四名人类专家。它还能够识别出与组织病理学解释一致的 OCM 图像中的形态学特征。
无标记 OCM 成像与基于深度学习的 CADx 方法相结合，有望在临床上用于有效筛查和诊断宫颈疾病。
