摘要： Purpose Urothelial carcinoma of the bladder (UCB) is the most common urinary cancer. White-light cystoscopy (WLC) forms the corner stone for the diagnosis of UCB. However, histopathological assessment is required for adjuvant treatment selection. Probe-based confocal laser endomicroscopy (pCLE) enables the visualization of the micro-architecture of bladder lesions during WLC, which allows for real-time tissue differentiation and grading of UCB. To improve the diagnostic process of UCB, computer aided classification of pCLE videos of in vivo bladder lesions were evaluated in this study. Materials and Methods We implemented pre-processing methods to optimize contrast and to reduce striping artifacts in each individual pCLE frame. Subsequently, a semi-automatic frame selection was performed. The selected frames were used to train a feature extractor, based on pre-trained ImageNet networks. A recurrent neural network, in specific long-short term memory (LSTM), was used to predict the grade of the bladder lesions. The differentiation of lesions was performed at two levels, namely (i) healthy and benign versus malignant tissue and (ii) low-grade versus high-grade papillary UCB. A total of 53 patients with 72 lesions were included in this study, resulting in approximately 140.000 pCLE frames. Results The semi-automated frame selection reduced the number of frames to approximately 66.500 informative frames. The accuracy for the differentiation of (i) healthy and benign versus malignant urothelium was 79% and (ii) high-grade and low-grade papillary UCB with 82%. Conclusions A feature extractor in combination with a LSTM results in an proper stratification of pCLE videos of in vivo bladder lesions.