Wu J.-WChang K.-YLI-CHEN FU2022-04-252022-04-25202100189456https://www.scopus.com/inward/record.uri?eid=2-s2.0-85121354503&doi=10.1109%2fTIM.2021.3134324&partnerID=40&md5=cc2989077a04e112ee583691bfbef719https://scholars.lib.ntu.edu.tw/handle/123456789/607427Confocal laser scanning microscopy (CLSM) is a non-destructive optical measurement system of high precision, applicable to the construction of three-dimensional topographies of biological cells and engineered materials at the micro- and sub-micro scales. Compressive sensing (CS) has recently been applied in microscope systems to reduce the amount of sampled data required for the reconstruction of images; however, the iterative nature of the CS recovery algorithm imposes high computational complexity. This paper presents an end-to-end non-iterative deep residual convolutional neural network (CNN) applicable to CLSM systems for CS-based reconstruction. In experiments and numerical simulations, the proposed scheme outperformed existing CS recovery algorithms in terms of reconstructed image quality as well as computation time. The proposed algorithm also enabled the reconstruction of images using samples obtained in different regions of an image at various sampling rates to overcome non-uniform information density. The reconstruction performance of the model in terms of robustness and efficiency were validated using real-world CLSM data obtained via random scanning patterns. IEEECompressed sensingCompressive sensingConfocal laser scanning microscopyConvolutional neural networkConvolutional neural networksCurrent measurementDeep residualEstimationImage qualityImage reconstructionTemperature measurementTrainingAdaptive opticsConfocal microscopyConvolutionDeep neural networksImage registrationInverse problemsIterative methodsRecoveryScanningSurface analysisCompressed-SensingConfocal-laser-scanning-microscopyImages reconstructionRecovery algorithmsUnder-samplingjournal article10.1109/TIM.2021.31343242-s2.0-85121354503