Objective To observe the value of multi-task improved nnU-Net model based on enhanced CT for segmenting primary oral cancer and predicting patients'relapse free survival(RFS).Methods Enhanced CT data of 186 cases of primary oral cancer were retrospectively analyzed,and a multi-task improved nnU-Net model was constructed for tumor segmentation and survival prediction tasks.Pre-training of tumor segmentation was completed with nnU-Net as the baseline network,and the accuracy of recognizing and segmenting tumor was improved by enhancing the decoder through the modified skip connection.Then univariable and multivariable regression analyses were used to select clinical features closely associated with RFS.Radiomics and deep learning features were also extracted to construct a survival prediction model,with fine-tuning of the above model.The training set,validation set and test set were divided at a ratio of 7∶2∶1.Dice similarity coefficient(DSC)was used to evaluate the segmentation performance of the modified model,and the consistency index C-index was used to verify the performance of the improved model for predicting RFS.Results DSC of the multi-task improved nnU-Net model(0.78)for segmenting primary oral cancer was superior to that of 3D Inception ResNet(0.65),3D InceptSENet(0.75)and 3D U-Net models(0.69),respectively,its C-index for predicting RFS(0.798)was higher than that of Cox regression model(0.744),ICARE model(0.761),random forest model(0.744),DeepSurv model(0.735),nnU-Net model(0.760)and radiology+nnU-Net model(0.744),respectively.DSC for segmenting primary oral cancer and C-index for predicting RFS of multi-task improved nnU-Net model were both superior to those of simple baseline network(0.653 and 0.649),baseline network+multi-scale convolution fusion(0.755 and 0.752),as well as baseline network combined with clinical features(0.764 and 0.759),radiomics features(0.770 and 0.764)and clinical+radiomics features(0.773 and 0.761),respectively.Conclusion Multi-task improved nnU-Net model could be used to effectively improve the accuracy of tumor segmentation and predicting patients'RFS.

Objective To observe the value of multi-task improved nnU-Net model based on enhanced CT for segmenting primary oral cancer and predicting patients'relapse free survival(RFS).Methods Enhanced CT data of 186 cases of primary oral cancer were retrospectively analyzed,and a multi-task improved nnU-Net model was constructed for tumor segmentation and survival prediction tasks.Pre-training of tumor segmentation was completed with nnU-Net as the baseline network,and the accuracy of recognizing and segmenting tumor was improved by enhancing the decoder through the modified skip connection.Then univariable and multivariable regression analyses were used to select clinical features closely associated with RFS.Radiomics and deep learning features were also extracted to construct a survival prediction model,with fine-tuning of the above model.The training set,validation set and test set were divided at a ratio of 7∶2∶1.Dice similarity coefficient(DSC)was used to evaluate the segmentation performance of the modified model,and the consistency index C-index was used to verify the performance of the improved model for predicting RFS.Results DSC of the multi-task improved nnU-Net model(0.78)for segmenting primary oral cancer was superior to that of 3D Inception ResNet(0.65),3D InceptSENet(0.75)and 3D U-Net models(0.69),respectively,its C-index for predicting RFS(0.798)was higher than that of Cox regression model(0.744),ICARE model(0.761),random forest model(0.744),DeepSurv model(0.735),nnU-Net model(0.760)and radiology+nnU-Net model(0.744),respectively.DSC for segmenting primary oral cancer and C-index for predicting RFS of multi-task improved nnU-Net model were both superior to those of simple baseline network(0.653 and 0.649),baseline network+multi-scale convolution fusion(0.755 and 0.752),as well as baseline network combined with clinical features(0.764 and 0.759),radiomics features(0.770 and 0.764)and clinical+radiomics features(0.773 and 0.761),respectively.Conclusion Multi-task improved nnU-Net model could be used to effectively improve the accuracy of tumor segmentation and predicting patients'RFS.

Objective:To compare the consistency in diagnosing and staging acute kidney injury (AKI) in children with chronic kidney disease (CKD) according to three criterias.Methods:Children with CKD hospitalized in the First Affiliated Hospital of Sun Yat sen University from January 2013 to December 2019 were analyzed retrospectively. These patients underwent serum creatinine examination more than twice during hospitalization. The AKI diagnosis and staging were performed for each patient according to the 2007 pRIFLE, 2012 KDIGO and 2018 pROCK criteria respectively. All the children were followed up for 1 year after discharge through outpatient visit, re-hospitalization or online consultation. The clinical characteristics and prognosis of CKD children with or without AKI that were diagnosed by 3 criteria were compared. Analysis of variance and chi-squared tests were used for the comparison among groups. Concordance between the different diagnostic criteria was evaluated using Cohen′s kappa coefficient.Result:A total of 2 551 children with CKD were included in this study, with an age of (8±4) years. There were 1 628 boys and 923 girls. Nephrotic syndrome was the most prevalent primary disease (55.4%), followed by lupus nephritis (11.2%) and purpura nephritis (8.2%). Among all stages of CKD, CKD category G1 was the most common type (2 146 cases, 84.1%), followed by CKD category G2 (221 cases, 8.7%). AKI occurence rates according to pRIFLE, KDIGO and pROCK criteria were 33.9% (866/2 551), 26.2%(669/2 551) and 19.5% (498/2 551) respectively (χ2=136.3, P<0.01). The diagnostic consistency within three criteria for AKI was high in children with CKD ( κ=0.702), but AKI staging consistency was low ( κ=0.329). Both the diagnosis and staging consistency of three AKI criteria were poor in children with CKD category G5 (all κ<0.400). The length of hospital stay (LOS), hospitalization costs, the occurence of intensive care unit (ICU) admission and in-hospital mortality were significantly higher in children with AKI diagnosed by different criteria ( P<0.05). After 1-year follow-up, the repeated admission rate and CKD staging progress significantly increased in children with AKI ( P<0.05). In children with baseline serum creatinine≥200 μmol/L, compared with children who did not experience AKI during hospitalization, the LOS and the hospitalization costs in children who were diagnosed AKI according to pRIFLE or pROCK criteria was significantly higher ( P<0.05). However, there was no significant difference in the LOS and hospitalization costs between children with or without AKI who were diagnosed according to KDIGO criteria (all P>0.05). Conclusions:AKI diagnosed by all of the three criteria (pRIFLE, KDIGO and pROCK criteria) was associated with the poor prognosis in children with CKD. However, in those whose baseline serum creatinine≥ 200 μmol/L, AKI diagnosed by pRIFLE and pROCK criteria could better reflect the poor outcomes than by KDIGO criteria.