Rodríguez Luis, Alva Alicia, Coronel Jorge, Caviedes Luz, Mendoza-Ticona Alberto, Gilman Robert, Sheen Patricia, Zimic Mirko
Centro de Excelencia en Tuberculosis Luz Caviedes Rojas, Hospital Regional Docente de Trujillo, La Libertad, Perú
Laboratorio de Investigación en Enfermedades Infecciosas, Laboratorio de Investigación y Desarrollo, Facultad de Ciencias y Filosofía, Universidad Peruana Cayetano Heredia, Lima, Perú
Rev Peru Med Exp Salud Publica. 2014 Jul-Sep;31(3):445-53.
To implement a system for remote diagnosis of tuberculosis and multidrug resistance (MDR) using the Microscopic-Observation Drug Susceptibility Assay (MODS) method in the Mycobacteria Laboratory, Trujillo Center of Excellence in Tuberculosis (CENEX-Trujillo). The system included a variant of an algorithm for recognition of Mycobacterium tuberculosis recently reported from digital images of MODS cultures of sputum samples.
The recognition algorithm was optimized using a retraining statistical model based on digital images of MODS cultures from CENEX-Trujillo. Images of 50 positive MODS cultures of patients with suspected multidrug-resistant tuberculosis were obtained between January and October 2012 in the CENEX-Trujillo.
The sensitivity and specificity to recognize strings of tuberculosis were 92.04% and 94.93% respectively using objects. The sensitivity and specificity to determine a positive tuberculosis field were 95.4% and 98.07% respectively using pictures.
The results demonstrated the feasibility of the implementation of telediagnostics in remote locations, which may contribute to the early detection of multidrug-resistant tuberculosis by MODS method.
在特鲁希略结核病卓越中心（CENEX - 特鲁希略）的分枝杆菌实验室中，使用显微镜观察药物敏感性试验（MODS）方法实施结核病和耐多药（MDR）远程诊断系统。该系统包括一种算法的变体，用于从痰标本MODS培养物的数字图像中识别结核分枝杆菌，该算法是最近报道的。
使用基于CENEX - 特鲁希略MODS培养物数字图像的再训练统计模型对识别算法进行优化。2012年1月至10月期间，在CENEX - 特鲁希略获取了50例疑似耐多药结核病患者的MODS阳性培养物图像。
使用物体识别结核链的敏感性和特异性分别为92.04%和94.93%。使用图片确定结核阳性区域的敏感性和特异性分别为95.4%和98.07%。
结果证明了在偏远地区实施远程诊断的可行性，这可能有助于通过MODS方法早期检测耐多药结核病。
