Cervical cancer disproportionally affects women in low- and middle-income countries, in part due to the difficulty of implementing existing cervical cancer screening and diagnostic technologies in low-resource settings. Single-board computers offer a low-cost alternative to provide computational support for automated point-of-care technologies. Here we demonstrate two new devices for cervical cancer prevention that use a single-board computer: 1) a low-cost imaging system for real-time detection of cervical precancer and 2) a low-cost reader for real-time interpretation of lateral flow-based molecular tests to detect cervical cancer biomarkers. Using a Raspberry Pi computer to provide real-time image collection and processing, we developed: 1) a low-cost, portable high-resolution microendoscope system (PiHRME); and 2) a low-cost automatic lateral flow test reader (PiReader). The PiHRME acquired high-resolution ( $4.4~\mu \text{m}$ ) images of the cervix at half the cost of existing high-resolution microendoscope systems; image analysis algorithms based on convolutional neural networks were implemented to provide real-time image interpretation. The PiReader acquired and analyzed images of a point-of-care human papillomavirus (HPV) serology test with the same contrast and accuracy as a standard flatbed high-resolution scanner coupled to a laptop computer, for less than one-fifth of the cost. Raspberry Pi single-board computers provide a low-cost means to implement point-of-care tools with automatic image analysis. This work demonstrates the promise of single-board computers to develop and translate low-cost, point-of-care technologies for use in low-resource settings.

中文翻译：

---

开发基于单板计算机的低成本宫颈癌预防护理点技术


宫颈癌对低收入和中等收入国家的妇女影响尤为严重，部分原因是在资源匮乏的地区实施现有的宫颈癌筛查和诊断技术存在困难。单板计算机提供了一种低成本的替代方案，为自动化护理点技术提供计算支持。在这里，我们展示了两种使用单板计算机预防宫颈癌的新设备：1）用于实时检测宫颈癌前病变的低成本成像系统，2）用于实时解释侧向血流的低成本读取器基于分子测试来检测宫颈癌生物标志物。使用Raspberry Pi计算机提供实时图像采集和处理，我们开发了：1）低成本、便携式高分辨率显微内窥镜系统（PiHRME）； 2) 低成本自动层析测试阅读器（PiReader）。 PiHRME 以现有高分辨率显微内窥镜系统一半的成本获取高分辨率 ( $4.4~\mu \text{m}$ ) 子宫颈图像；实施基于卷积神经网络的图像分析算法以提供实时图像解释。 PiReader 采集并分析了即时人乳头瘤病毒 (HPV) 血清学检测图像，其对比度和准确性与连接笔记本电脑的标准平板高分辨率扫描仪相同，而成本不到五分之一。 Raspberry Pi 单板计算机提供了一种低成本的方法来实施具有自动图像分析功能的护理点工具。这项工作展示了单板计算机在开发和转化低成本、即时护理技术以在资源匮乏的环境中使用的前景。