首页|Smartphone-based, automated detection of urine albumin using deep learning approach
Smartphone-based, automated detection of urine albumin using deep learning approach
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NSTL
Elsevier
Detection of microalbuminuria is a vital factor to prevent the progression of cardiovascular and renal disease. Several clinical studies on large population has shown the significance of dipstick in detection of micro-albuminuria. Although poor-quality of visual assessment has hindered its clinical utility. Automatic detection of trace and higher proteinuria is critical for detection of asymptomatic individuals. In this work, an automated, accessory free analytical system using smartphone for quantification of albuminuria has been investigated. A customised convolutional neural network (CNN) model along with different color spaces has been used to classify albumin concentration in urine dipstick. To mitigate ambient light conditions, smartphone camera was used in "Flash ON " mode. Performance of CNN model in different lighting conditions and with different smartphone models was studied and an accuracy of 88% was achieved on test data.
Deep learningUrine albuminSmartphonePoint -of -care testingDipstickCOLORIMETRIC DETECTIONPROTEINURIADIPSTICKANALYZERRISK
NSTL
Elsevier
