Due to the low contrast of CT images,irregular shape of the liver,and blurred boundaries with adjacent organs,the existing methods based on convolutional neural network underperform in liver segmentation tasks,especially for boundary recognition and small object detection.A novel liver segmentation method is proposed based on multi-scale feature fusion and attention,namely MFFA UNet.Multi-scale feature fusion is firstly employed to acquire abundant segmentation details,while spatial and channel attention mechanisms are utilized to capture global spatial and inter-channel relationships.Additionally,a deep supervision module fully leverages the output of intermediate hidden layers,enhancing the learning capability of the network,which in turn accelerates the network's convergence speed.Moreover,a hybrid loss function is adopted to address the issue of class imbalance,further boosting the model's segmentation efficacy.Experimental results demonstrate that the proposed MFFA UNet outperforms the prevailing segmentation networks on the public LITS dataset,producing results that are closer to the ground truth.

Objective To investigate the prevalence and severity of symptoms and to construct symptom networks in head and neck cancer patients during treatment to identify core symptoms and symptom clusters.Methods 366 patients who were hospitalized in 3 tertiary hospitals in Changsha were selected using convenience sampling from March to October 2022 and asked to complete the M.D.Anderson Symptom Inventory-Head & Neck.Exploratory factors analysis was used to extract the symptom clusters,and R packages were used to construct the symptom severity network and symptom clusters network.The centrality indexes of the networks,including strength,closeness,and betweenness,were analyzed to identify core symptoms and core symptom cluster.Results The most common symptoms in head and neck cancer patients during treatment were dry mouth(93.44％),fatigue(89.07％),loss of appetite(86.34％),and difficulty swallowing or chewing(85.79％),and the most severe symptoms were dry mouth,loss of appetite,oral or pharyngeal mucus,and difficulty swallowing or chewing.4 symptom clusters were extracted,namely oral-pharyngeal,gastrointestinal,emotional-sleep,and sickness-sensing behavioral,which could explain 67.415％of the total variance.In the symptom severity network,oral or pharyngeal mucus(rs=9.60)was a symptom with the highest strength.In the symptom clusters network,oral or pharyngeal mucus(rs=1.20),nausea(rs=1.00),fatigue(rs=1.10),and drowsiness(rs=0.97)were the symptoms with the highest strength across 4 symptom clusters.Conclusion Oral or pharyngeal mucus,nausea,fatigue,and drowsiness are the core symptoms of symptom clusters in head and neck cancer patients during treatment.Oral-pharyngeal symptom cluster is the core symptom cluster.It is recommended that clinical staff should develop interventions based on the core symptoms and symptom cluster to implement precise symptom management and improve symptom management efficiency.

OBJECTIVE To investigate the predictive value of prognostic nutritional index(PNI) and lymphocyte-monocyte ratio(LMR) in severe radiotherapy-induced oral mucositis(RIOM) during treatment of patients with head and neck cancer,and to construct a risk prediction model and test the prediction effect. METHODS A total of 502 patients with head and neck cancer who underwent radiotherapy were recruited from September 2021 to October 2023 in Xiangya Hospital Central South University. The participants were randomly divided into training group and validation group at a ratio of 7:3. According to whether severe RIOM occurred,they were divided into severe RIOM group and non-severe RIOM group. Univariate analysis and logistic regression analysis were used to screen the risk factors of severe RIOM. The receiver operating characteristic(ROC) curve was used to evaluate its prediction effect and R4.3.2 software was used to draw nomograms and decision curve. RESULTS The risk prediction model for patients with head and neck cancer during treatment had five factors,including the number of comorbidities(OR=2.221,95％CI=1.185-4.165),surgical history(OR=2.938,95％CI=1.393-6.198),the degree of tumor differentiation(OR=1.511,95％CI=1.090-2.094),PNI(OR=0.892,95％CI=0.852-0.934),LMR(OR=0.512,95％CI=0.254-1.030). Model formula:Y=2.102+0.413×degree of differentiation+0.798×number of comorbidities+1.078×surgical history-0.114×PNI-0.669×LMR. The validation results of the prediction model showed that the area under the ROC curve of the training group was 0.847(P<0.001),the area under the curve of the validation group was 0.808(P<0.001),and the P values of the Hosmer-Lemeshow test of the modeling group and the validation group were both greater than 0.05. The decision curve was above the reference line within most of the high-risk thresholds. CONCLUSION The risk prediction model constructed in this study has good effect,which can predict the risk of severe RIOM during radiotherapy in patients with head and neck cancer,providing the reference for taking preventive intervention measures for high-risk patients.