Abstract. Neural networks have recently been proposed for multi-label classification because they are able to capture and model label dependencies in the output layer. In this work, we investigate limitations of BP-MLL, a neural network (NN) architecture that aims at minimizing pairwise ranking error. Instead, we propose to use a comparably simple NN approach with recently proposed learning techniques for large-scale multi-label text classification tasks. In particular, we show that BP-MLL's ranking loss minimization can be efficiently and effectively replaced with the commonly used cross entropy error function, and demonstrate that several advances in neural network training that have been developed in the realm of deep learning can be effectively employed in this setting. Our experimental results show that simple NN models equipped with advanced techniques such as rectified linear units, dropout, and AdaGrad perform as well as or even outperform state-of-the-art approaches on six large-scale textual datasets with diverse characteristics.
PurposeThe authors evaluate the prevalence of malnutrition and its effect on the postoperative morbidity of patients after surgery for colorectal cancer.MethodsThree hundred fifty-two patients were enrolled prospectively. Nutritional risk screening 2002 (NRS 2002) score was calculated through interview with patient on admission. Clinical characteristics, tumor status and surgical procedure were recorded.ResultsThe prevalence of patients at nutritional risk was 28.1 per cent according to the NRS 2002. The rate of postoperative complication was 27%. There was a significant difference in postoperative complication rates between patients at nutritional risk and those not at risk (37.4% vs. 22.9%, P = 0.006). Nutritional risk was identified as an independent predictor of postoperative complications (odds ratio, 3.05; P = 0.045). Nutritional risk increased the rate of anastomotic leakage (P = 0.027) and wound infection (P = 0.01).ConclusionNRS may be a prognostic factor for postoperative complication after surgery for colorectal cancer. A large scaled prospective study is needed to confirm whether supplementing nutritional deficits reduces postoperative complication rates.
Purpose: To investigate the usefulness of apparent diffusion coefficient (ADC) values derived from histogram analysis of the whole rectal cancer as a quantitative parameter to evaluate pathologic complete response (pCR) on preoperative magnetic resonance imaging (MRI). Materials and Methods: We enrolled a total of 86 consecutive patients who had undergone surgery for rectal cancer after neoadjuvant chemoradiotherapy (CRT) at our institution between July 2012 and November 2014. Two radiologists who were blinded to the final pathological results reviewed post-CRT MRI to evaluate tumor stage. Quantitative image analysis was performed using T 2 -weighted and diffusion-weighted images independently by two radiologists using dedicated software that performed histogram analysis to assess the distribution of ADC in the whole tumor. Results: After surgery, 16 patients were confirmed to have achieved pCR (18.6%). All parameters from pre-and post-CRT ADC histogram showed good or excellent agreement between two readers. The minimum, 10th, 25th, 50th, and 75th percentile and mean ADC from post-CRT ADC histogram were significantly higher in the pCR group than in the non-pCR group for both readers. The 25th percentile value from ADC histogram in post-CRT MRI had the best diagnostic performance for detecting pCR, with an area under the receiver operating characteristic curve of 0.796. Conclusion: Low percentile values derived from the ADC histogram analysis of rectal cancer on MRI after CRT showed a significant difference between pCR and non-pCR groups, demonstrating the utility of the ADC value as a quantitative and objective marker to evaluate complete pathologic response to preoperative CRT in rectal cancer.
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.