In this paper, we present a generic deep convolutional neural network (DCNN) for multi-class image segmentation. It is based on a well-established supervised end-to-end DCNN model, known as U-net. U-net is firstly modified by adding widely used batch normalization and residual block (named as BRUnet) to improve the efficiency of model training. Based on BRU-net, we further introduce a dynamically weighted cross-entropy loss function. The weighting scheme is calculated based on the pixel-wise prediction accuracy during the training process. Assigning higher weights to pixels with lower segmentation accuracies enables the network to learn more from poorly predicted image regions. Our method is named as feedback weighted U-net (FU-net). We have evaluated our method based on T1-weighted brain MRI for the segmentation of midbrain and substantia nigra, where the number of pixels in each class is extremely unbalanced to each other. Based on the dice coefficient measurement, our proposed FU-net has outperformed BRU-net and U-net with statistical significance, especially when only a small number of training examples are available. The code is publicly available in GitHub 1 .
590 Background: We hereby evaluated the histopathological and radiological alterations of tumor characteristics after receiving NACT and the clinical significance of the changes of adjuvant therapy based on these findings. Methods: A pathological assessment of tumor features including ER, PR, HER2 and proliferation markers (Ki67 and SPAG5) status in pre and post NACT tumors tissue have been centrally evaluated in two cohorts [Nottingham University Hospital (NUH; n=850) and Australian cohort (n=250 patients)]. Since 2013 any change in the ER and HER2 status from negative (-) [in the pre NACT biopsies] to positive (+) [in the post NACT surgical specimens] received additional adjuvant therapy (Endocrine therapy (ET) for ER+ and Trastuzumab for HER2+ cases) in NUH. MRI volumetric and texture changes have been assessed in 400 cases. The primary end point was disease free survival (DFS; median follow-up = 62 months). Results: 10% of pre NACT HER2- cases had been converted to post NACT HER2+ and those cases who subsequently received adjuvant Trastuzumab had achieved 92% 5-year DFS compared to those who remained HER- in post NACT specimens (58% 5-year DFS); (HR (95% CI)= 0.25 (0.08-0.80); p=0.016). While 13% of pre NACT HER2+ tumors were converted into HER2- in post NACT surgical specimens and had similar 5-year DFS to those who remained post NACT HER2+ (5-year DFS= 94% vs., 87%; p=0.613). Loss of PR in the residual disease of pre NACT ER+ BC was associated with shorter 5-year DFS after ET compared to those who remained post NACT PR+ (HR (95% CI)=2.1 (1.25-3.46); p=0.005). After NACT, 40% of pre NACT SPAG5+ cases were converted into post NACT SPAG5- and these patients had prolonged DFS compared to those who remained SPAG5+ in post NACT specimens (27%) [5-year DFS=84% vs 49%; (HR (95% CI)= 3.8 (2.1-6.9); p<0.0001). A prognostic model has been generated including factors in table (AUC = 0.854 (95% CI) = 0.777-.0.931; p= 0.00000001]. Conclusions: We hereby showed evidences a change of treatment strategy based on the changes in the tumor post NACT phenotype gives the optimal choice of treatment eg., the introduction of HER2 targeting therapy for the conversion of HER2– to HER2+ phenotype after NACT improved DFS. Multivariate Cox regression model for 5-year DFS. [Table: see text]
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.