Objective:To assess the survival-predictive value of TNM and Lugano staging systems in patients with primary gastro-intestinal lymphoma (PGL). Methods:A total of 73 patients with PGL were recruited from February 2001 to August 2013. All patients were diagnosed according to the TNM and Lugano staging systems. Five-year survival rate was used as the major clinical outcome. Sur-vival curves were plotted using the Kaplan–Meier method and analyzed with the log-rank test. The prognostic value of different vari-ables for clinical outcomes was assessed using the Cox multiple regression model. Results:The median follow-up time of surviving pa-tients was 42.4 months (range:1.3-158.6 months). The estimated 5-year overall survival rate was 77.82%. When diagnosed with the TNM system, the 5-year survival rates in stagesⅠ,Ⅱ,Ⅲ, andⅣwere 100%, 90.0%, 67.4%, and 22.2%, respectively (χ2=17.7956, P=0.0005). When staged by the Lugano system, the 5-year survival rates in stagesⅠ,Ⅱ,ⅡE , andⅣwere 100%, 100%, 70.7%, and 46.2%, respectively (χ2=15.6776, P=0.0013). Cox analysis showed that the invasion depth (T) (P=0.0181) and metastasis (M) (P=0.0031) were covariates that were prognostically significant for the overall survival. Conclusion:The TNM staging system is more ac-curate than the Lugano system in predicting the 5-year survival rate of patients with PGL.

Deep learning (DL) has achieved state-of-the-art performance in many digital pathology analysis tasks. Traditional methods usually require hand-crafted domain-specific features, and DL methods can learn representations without manually designed features. In terms of feature extraction, DL approaches are less labor intensive compared with conventional machine learning methods. In this paper, we comprehensively summarize recent DL-based image analysis studies in histopathology, including different tasks (e.g., classification, semantic segmentation, detection, and instance segmentation) and various applications (e.g., stain normalization, cell/gland/region structure analysis). DL methods can provide consistent and accurate outcomes. DL is a promising tool to assist pathologists in clinical diagnosis.