Predicting 5-Year Survival in Gastric Cancer Patients Using Iliopsoas Muscle CT Radiomics and Machine Learning Techniques

Predicting 5-Year Survival in Gastric Cancer Patients Using Iliopsoas Muscle CT Radiomics and Machine Learning Techniques | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Predicting 5-Year Survival in Gastric Cancer Patients Using Iliopsoas Muscle CT Radiomics and Machine Learning Techniques Yuan Hong, Yifan Li, Peng Zhang, Haosong Chen, Yixian Chen, Yang Yu, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5350805/v1 This work is licensed under a CC BY 4.0 License Status: Under Revision Version 1 posted 4 You are reading this latest preprint version Abstract Objectives Sarcopenia, linked to postoperative survival in cancer patients, was investigated in this study. The research explored the relationship between CT imaging features of muscles in gastric cancer patients and their survival. Additionally, the study aimed to create a quantifiable survival prediction model using artificial intelligence. Methods In a retrospective study, 100 patients who underwent radical gastrectomy for gastric cancer were analyzed. After identifying sarcopenia using the psoas muscle index, clinical factors related to patient survival were investigated. Imaging features were extracted from manually delineated iliopsoas muscles and used in 11 machine learning algorithms. After completing the model training, we used a dataset comprising 34 patients from a secondary center as an external validation set to evaluate the model’s classification performance. After identifying the optimal model, we further explored the fusion methods of clinical omics and radiomics. Based on this, we constructed a predictive model for estimating the five-year survival rate of patients. Results Clinical survival analysis highlighted age and tumor M stage as relevant factors. For the task of predicting five-year survival, we found that the Logistic Regression (LR) model without clinical feature fusion exhibited the most balanced and superior performance. Specifically, the AUC (Area Under Curve) values of this model on the training set, internal validation set, and external validation set were 0.82, 0.72, and 0.69, respectively. Additionally, the model’s accuracy remained relatively stable, approximately around 70%. Conclusions In this study, we developed a machine learning model based on preoperative CT imaging data of gastric cancer patients to predict their five-year survival rate. The model can achieve about 70% accuracy. Additionally, we explored the necessity and rationale of incorporating clinical independent factors into this predictive model. The results indicated a significant correlation between muscle imaging features and overall patient survival, highlighting the importance of sarcopenia in the clinical management of gastric cancer patients. sarcopenia CT radiomics machine learning prediction gastric cancer Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 1. Introduction Sarcopenia, initially proposed by Irwin Rosenberg in the 1980s[ 1 ], refers to age-related muscle loss. Over time, its definition has evolved to include degenerative changes in skeletal muscle mass, strength, and function. In 2022, the International Sarcopenia Working Group provided a detailed definition, describing it as a progressive reduction in muscle mass, strength, or physiological function associated with aging.[ 2 ] The European and Asian Working Groups played crucial roles in standardizing diagnostic criteria, which now encompass assessments of muscle strength, mass, and physical performance.[ 3 ] A meta-analysis of 38 studies (7,843 patients) revealed that low skeletal muscle index (SMI) at diagnosis predicts poor survival, especially in gastric and esophageal cancer patients.[ 4 ] Gastric cancer patients have a higher sarcopenia incidence, linked to age, gender, nutritional status, and disease stage.[ 5 ] Gastric cancer-induced metabolic changes, including imbalances in carbohydrate, protein, and fat metabolism, contribute to sarcopenia.[ 6 ]Given that the diagnosis of sarcopenia primarily relies on the cross-sectional data of the psoas major muscle, and previous research on sarcopenia has commonly focused on the iliac psoas muscle as the target muscle, this study also follows this research path. The iliac psoas muscle was considered the core research subject, and precise delineation was performed using two-dimensional sections and three-dimensional models.[ 1 – 6 ] Despite the link between sarcopenia and gastric cancer, research on using sarcopenia to reveal clinical information remains limited. Regional variations in diagnostic standards, traditional muscle mass assessment limitations, and insufficient clinical awareness contribute to this gap.[ 7 ] Radiomics, an emerging technology, extracts multidimensional features from medical images (such as CT and MRI) using automated algorithms. It supports precise disease diagnosis, classification, staging, and prognosis prediction. In sarcopenia research, radiomics aims for intelligent, accurate diagnosis and direct application in assessing patient survival and prognosis.[ 8 ] However, overall research in this field remains limited, and the application of AI technology requires further expansion.[ 9 ] Our objective is to integrate AI algorithms for accurate prediction of 5-year survival in gastric cancer patients. Leveraging AI technology, we aim to enhance personalized treatment and prognosis management. 2. Material and Methods 2.1. Clinical Data Collection This study, approved by the Ethics Committee, retrospectively analyzed clinical data from 100 gastric cancer patients diagnosed between December 2018 and June 2019. The detailed inclusion and exclusion criteria for patients are shown in the supplementary materials. After radical gastrectomy, continuous follow-up was conducted. Of the initial sample, 12 patients declined participation, and 17 were lost to follow-up. However, five-year survival data for the remaining 71 patients were accurately recorded. To validate the performance of the model and mitigate the risk of overfitting, our center, upon obtaining authorization, utilized preoperative CT and clinical data from 34 patients diagnosed with gastric cancer who underwent major gastric surgery at Hanshan County People’s Hospital between 2017 and 2018 as an external validation set. The inclusion criteria for this validation set were consistent with our center’s internal standards to ensure data comparability. All patients completed a 5-year follow-up, during which survival data were collected. The CT dataset for this study was obtained from routine preoperative scans using a 256-channel scanner (Philips Brilliance i CT). Image acquisition utilized a 512 × 512 matrix, maintaining in-plane resolution between 0.62 × 0.62 mm and 0.86 × 0.86 mm. To enhance image visualization, 3D Slicer software adjusted the CT image window width (WW) to 149 and window level (WL) to 40, with a threshold range of -35 to 115.[ 10 ] 2.2. Assessment of Sarcopenia In the current research on sarcopenia diagnosis, most studies prefer using non-contrast computed tomography (CT) with specific abdominal window settings for image analysis. This method effectively distinguishes muscle tissue from surrounding structures, providing detailed muscle imaging. It significantly enhances the accuracy of muscle mass and quality assessment. Our study also employs this standard CT imaging technique for detailed muscle morphology analysis.[ 11 – 14 ] The gold standard for diagnosing sarcopenia involves CT imaging, specifically measuring skeletal muscle area at the L3 vertebral level and calculating the SMI.[ 15 , 16 ]A newly introduced method uses bioelectrical impedance analysis (BIA) to assess muscle content.[ 17 , 18 ] The psoas muscle index (TPI) is also valuable, standardized by measuring psoas muscle area at L3.[ 19 ]TPI assessments are more practical than BIA, which can be influenced by body water content. [ 20 ]Our study adheres to TPI standards, considering clinical examination results and dual criteria (6-meter gait speed and grip strength) for high-risk sarcopenia patients. Figure 1 outlines the psoas muscle delineation process at the bilateral L3 plane, ensuring reproducibility and study result accuracy. To diagnose sarcopenia, we adhere to international consensus standards for cancer patients.[ 21 ] Sarcopenia is diagnosed when TPI falls below 385 mm²/m² for females or 545 mm²/m² for males.[ 22 ] We used the lasso tool in 3D Slicer 5.4.0 software to outline 2D images, with collaborative input from experienced radiologists and final review by a senior physician to ensure measurement accuracy and reliability. 2.3. Clinical Data Statistical Analysis In this study, we diagnosed sarcopenia using TPI at the L3 level, dividing patients into sarcopenia and non-sarcopenia groups. We collected 11 clinical indicators, including height, weight, and TNM staging, for the training set and conducted statistical analysis. The external validation set, limited by data authorization, only includes basic pathological and clinical information, so no in-depth analysis was performed. Correlation analysis revealed no significant association between clinical factors and sarcopenia diagnosis in training set. To explore survival factors, we used Kaplan-Meier survival analysis and univariate COX regression. Multivariate COX regression identified significant survival factors. Combining these with radiomics features, we trained and tested machine learning models, ultimately selecting an accurate 5-year survival prediction model for gastric cancer patients. To ensure the validity of the analysis results, the research team conducted a post-hoc power test on patient survival time after completing the statistical analysis. 2.4. Radiomics Workflow and Muscle Segmentation The overall radiomics workflow is shown in Fig. 2 . In the study, abdominal plain CT images were used. Researchers manually delineated the bilateral iliopsoas muscles using 3dsclier software with standard abdominal window settings. The specific 3D delineation process is illustrated in Fig. 3. To ensure accuracy and minimize human error, two radiologists independently performed the manual delineation, which was then reviewed and fine-tuned by a senior radiologist. We used the pyradiomics software package in Python to extract radiomic features from the 3D iliopsoas muscle images of the patients. Radiomic features were obtained using fixed bin number discretization (bin number = 5) and voxel resampling (X, Y, Z = 3 × 3 × 3 mm³).After feature extraction, we normalized features for subsequent data processing. We assessed feature correlation using the T-test and Spearman coefficient. Highly correlated features (coefficient > 0.9) were reduced to simplify model complexity. A Lasso regression model screened significant prediction-contributing features. The feature engineering process is detailed in Table 1 , and feature statistics are shown in Fig. 4 . Machine learning algorithms excel at handling large-scale, high-dimensional clinical information. They uncover complex patterns to achieve precise predictions, surpassing traditional statistical methods and enhancing disease diagnosis accuracy and efficiency. We innovatively apply these algorithms to prognosis modeling for sarcopenia patients, evaluating predictive performance across 11 selected algorithms to identify the optimal model for analyzing patient imaging features. Key features were input into 11 machine learning algorithms (SVM:support vector machine, KNN:k-nearest neighbors, RandomForest, ExtraTrees:extremely randomized trees, XGBoost, LightGBM:light gradient boosting machine, NaiveBayes: naive bayes classifier, AdaBoost:adaptive boosting, GradientBoosting, LR:logistic regression, MLP:multi-layer perceptron). We randomly split the patient dataset (8:2) for training and testing. After completing the model training phase, we evaluated the extrapolation performance of the constructed model and monitored the risk of overfitting using external validation set data. Based on the performance results on the validation set, we systematically screened the models and ultimately identified the best-performing model. We evaluated model performance using ROC curves on training and test sets, comparing predictive accuracy via AUC. In constructing the 5-year survival prediction model, we analyzed the impact of clinical and radiomic features, as well as their fusion, across different algorithm frameworks. We utilize the CLEAR[ 23 ] form and METRICS tool[ 24 ] to provide a comprehensive assessment of the quality of imaging work, please see the attached Supplemental Checklist 1 and Supplemental Checklist 2 for specific assessment details. Table 1 , Radiomics Feature Extraction Settings Setting Determination Bin Method FBN Bin Amount 5 Method SitkNearestNeighbor Resample Filter 1 Resample Spacing X 3mm Resample Spacing Y 3mm Resample Spacing Z 3mm FBN,fixed bin number 2.5. Statistics In the clinical data statistics phase, we securely saved patient information in anonymized Excel format. Statistical analysis used SPSS 27.0 for categorical variables (chi-square test) and continuous variables (t-test or Mann-Whitney U test). Survival analysis employed Kaplan-Meier (KM) and log-rank tests. Radiomics operations were implemented in Python (scikit-learn, pandas) within Jupyter Notebook. High-quality images were plotted using matplotlib. Radiomic features from imaging data were securely stored in Excel for subsequent analysis. 3. Results 3.1. Diagnosis of Sarcopenia Following Pelten et al.'s[ 25 ] criteria, we accurately located L3 vertebral plane images using 3D Slicer software. The lasso tool meticulously outlined bilateral psoas major muscles. We calculated TPI by dividing total muscle area by the square of patient height. Patients with TPI below 385 mm²/m² (females) or 545 mm²/m² (males) were diagnosed with sarcopenia. As shown in Table 2 , our training set included 44 sarcopenia patients (37 males, 7 females) and 27 non-sarcopenia patients (22 males, 5 females).The validation set (shown in Table 3 ) included 11 sarcopenia patients (9 males, 2 females) and 23 non-sarcopenia patients (20 males, 3 females). Notably, the sarcopenia group had significantly lower psoas major index than the non-sarcopenia group for both genders. 3.2. Clinical Baseline Variability Analysis We distinguished gastric cancer patients in training set into sarcopenia and non-sarcopenia groups, organizing 11 clinical data points (Table 4 ). The clinical baseline data of the validation set are presented in the Table 5 . It should be noted that due to the limitations of the clinical data in the validation set, this dataset was not used for in-depth differential analysis. Statistical methods varied by data type: chi-square tests (with continuity correction, Pearson, or Fisher’s exact) for binary variables (gender, surgical method, M stage), RC contingency table chi-square for multi-category variables (T stage, N stage), t-test for height (continuous), and Mann-Whitney U test for age, weight, BMI, hospital stay, and surgery time (continuous). Statistical analysis revealed that only surgery time significantly differed between sarcopenia and non-sarcopenia groups. Sarcopenia patients had longer surgery times. Other clinical variables showed no significant differences. This suggests that sarcopenia may complicate gastric cancer surgery, warranting further research into underlying mechanisms and impacts. 3.3. Clinical Survival Analysis We analyzed 5-year survival data using Kaplan-Meier (KM) survival analysis for sarcopenia and non-sarcopenia groups (Fig. 5 ). No significant difference in 5-year survival rates was observed. Correlation tests on clinical factors (Supplementary Table 1) revealed no significant association with sarcopenia. We performed univariate Cox regression analysis to assess the impact of various factors on patient survival. Age, weight, BMI, T stage, N stage, and M stage all significantly affected survival. Collinearity diagnostics confirmed independence among these factors, with variance inflation factors (VIF) below 5 (Tables 6 and 7 ). We incorporated six significant factors (age, weight, BMI, T stage, N stage, and M stage) into a multivariate Cox regression model. The results highlighted age and tumor M stage as independent and significant factors affecting the 5-year survival rate of gastric cancer patients (Table 8 ).A post-hoc power test was conducted on patient survival time, with a diagnostic efficiency of 0.834, confirming the validity of the statistical analysis. Detailed results are provided in Table 9 . 3.4. Predictive Modelling of 5-year Survival We transformed patients’ five-year survival status into binary data: ‘1’ for patients surviving more than five years and ‘0’ for those who passed away within the follow-up period. We selected radiological features and clinical variables as input, using 11 machine learning models. Comprehensive evaluation occurred via five-fold cross-validation.The classification performance and generalizability of each model is demonstrated in the external validation set. In this paper, we present the performance results of 11 models on both the training and test sets, as detailed in the Table 10 below. In terms of model accuracy, the external validation accuracy of both the Naive Bayes and K-Nearest Neighbors (KNN) models is below 0.5, indicating that the predictive performance of these two models is unacceptable. The ROC images of the two models are shown in Fig. 6 . Upon analysis, the KNN model exhibited excessive prediction bias and an abnormally high AUC, indicating poor generalization ability and insufficient predictive accuracy. Therefore, the research team decided to discard this model and explore other superior models. Furthermore, by analyzing the ROC curves of each model (as shown in the Fig. 7 ), we found that the AUC values of AdaBoost, Extra Trees, Gradient Boosting, Random Forest, Support Vector Machine (SVM), and XGBoost on the test set all exceeded 0.95, demonstrating extremely high predictive capabilities. However, considering the results on the validation set, the AUC values of the AdaBoost and Gradient Boosting models on the validation set were close to 0.5, indicating significant overfitting to the training set. Additionally, the performance of the Random Forest and Extra Trees models on the external validation set was also poor, raising doubts about their classification performance. For the SVM and XGBoost models, their accuracy and AUC values on the validation set were both stable around 0.7. The ROC images of the two models are shown in Fig. 8 .However, given the excessively high AUC values on the training set, these two models still present a significant risk of overfitting. Although they show potential for achieving high-quality classification models with larger training data volumes, due to the data limitations of this study, we currently do not recommend these two models as the optimal choices. Among the remaining three models, Logistic Regression (LR), LightGBM, and Multilayer Perceptron (MLP) exhibited stable AUC and accuracy on both the training and test sets, with relatively low risk of overfitting.The ROC images for the three sets of models are shown in Fig. 9 . By comprehensively comparing the classification performance of these three models, we found that the LR model maintained an accuracy and AUC value of around 0.7 on both the training and test sets, demonstrating good classification performance. Therefore, we consider the LR model to have performed the best in this study. Ultimately, we systematically integrated the collected clinical independent factors. Subsequently, to explore potential ways to enhance the model’s predictive ability, we combined these clinical features with imaging features. A comparison of the ROC curves of the two groups of models is shown in Fig. 10 and the model accuracy is detailed in Table 11 .However, analysis of the fused model revealed an interesting phenomenon: although the classification accuracy of this model improved on the internal test set, its performance on the external validation set significantly declined. Based on this observation, we cautiously concluded that while the introduction of clinical features unique to our center enhanced the model’s internal test classification performance, its generalization ability correspondingly weakened. As shown in Fig. 11 , analysis of the DCA curves for the two groups of models shows that incorporating clinical features enhances net benefits in the training and internal test sets. However, the performance in the external validation set is suboptimal, slightly lower than that of the non-integrated model. This indicates that the performance improvement from integrating clinical features is not consistently demonstrated in external validation. Therefore, in the specific context of this study, we evaluated that the LR model without clinical feature fusion exhibited the best performance, achieving approximately 70% accuracy on new datasets, providing valuable reference for our subsequent research. 4. Discussion Sarcopenia, characterized by the loss of skeletal muscle mass and function, is a critical factor influencing clinical outcomes in oncology patients. The diagnosis of sarcopenia has evolved with several mainstream methods being utilized. Commonly used screening tools include the five-item questionnaires SARC-F and SARC-CalF.[ 26 , 27 ] Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) are considered gold standards due to their accuracy in quantifying muscle mass[ 28 – 30 ], particularly the Skeletal Muscle Index (SMI) at the lumbar vertebra level[ 31 ]. Dual-energy X-ray Absorptiometry (DXA) is another widely used method, offering a balance between accuracy and accessibility[ 32 ]. Bioelectrical Impedance Analysis (BIA) provides a non-invasive and cost-effective alternative, though it is less precise compared to imaging techniques[ 33 , 34 ]. Early identification and management of sarcopenia are crucial, as it is associated with increased chemotherapy toxicity, postoperative complications, and reduced survival rates[ 31 , 32 ]. Therefore, integrating these diagnostic methods into routine clinical practice can significantly enhance patient care and treatment outcomes. Prognostic modeling in sarcopenia among oncology patients has advanced significantly in recent years. Current mainstream models incorporate both sarcopenia and systemic inflammation markers, such as the neutrophil/lymphocyte ratio (NLR) and lymphocyte/monocyte ratio (LMR), to predict clinical outcomes. Studies have shown that combining sarcopenia with low LMR is a sensitive prognostic factor for overall survival (OS) and progression-free survival (PFS) in head and neck cancer patients[ 29 ]. Additionally, sarcopenia has been identified as a negative predictive factor for survival outcomes and postoperative complications in cholangiocarcinoma[ 32 ]. These models underscore the importance of integrating muscle mass assessments with inflammatory markers to enhance prognostic accuracy in oncology patients[ 32 , 34 ]. In our clinical cohort, patients with sarcopenia exhibited significantly longer surgery times compared to those without sarcopenia, even when undergoing the same surgical method. This suggests a potential link between sarcopenia and malignant tumors. Although correlation analysis of clinical factors was conducted, no significant association with sarcopenia was found[ 35 ], possibly due to the limited sample size and limitations of traditional statistics. Additionally, in-depth analysis of 5-year survival data revealed the impact of patient age and tumor M stage on prognosis. An independent prediction model highlighted the key role of clinical factors in assessing long-term survival. However, after applying clinical modality fusion, we observed an unexpected decline in the model’s extrapolation performance. This phenomenon indirectly suggests that the independent influencing factors selected using our center’s data may have suffered from a certain degree of training set overfitting. Therefore, this research team does not recommend directly applying the selected clinical independent factors to new datasets. Regarding the actual clinical significance of the two selected clinical factors, we plan to conduct more in-depth clinical data collection and analysis to gain a more comprehensive understanding. This study applied muscle radiomics features from gastric cancer patients to predict medium- and long-term survival. The model demonstrated good classification performance in the classification tasks and exhibited stable extrapolation capability, thereby strongly proving the great potential of artificial intelligence technology in clinical applications. [ 36 , 37 ]Accurate prediction of overall survival time based on muscle mass is crucial for personalized treatment planning. This study acknowledges limitations due to a small sample size and single-center design, potentially affecting model accuracy. Efforts will focus on expanding the clinical cohort and collecting high-quality data. While current models rely on traditional machine learning omics algorithms, the research group aims to enhance performance by exploring advanced radiomics techniques, including Convolutional Neural Networks (CNNs). Expectations are high that AI technology will contribute to sarcopenia research in oncology. 5. Conclusions In this study, we developed a machine learning model based on preoperative CT imaging data of gastric cancer patients to predict their five-year survival rate. The model can achieve about 70% accuracy. Additionally, we explored the necessity and rationale of incorporating clinical independent factors into this predictive model. The results indicated a significant correlation between muscle imaging features and overall patient survival, highlighting the importance of sarcopenia in the clinical management of gastric cancer patients. This study not only provides new tools and methods for prognostic evaluation of gastric cancer patients but also emphasizes the significance of artificial intelligence technology in enhancing the quality of gastric cancer treatment decisions. Declarations Funding: This study was supported by a grant from the Graduate Student Research and Practice Innovation Program of Anhui Medical University (YJS20230080). Competing interests: The authors declare no potential conflict of interest in the research, writing, and publication of this article. Author Contribution Concept and design:Yuan Hong and Bo Chen; data collection and analysis:Yifan Li , Peng Zhang , Haosong Chen , Yixian Cheng , Zimo Zhang and Kang Cheng; drafting of the article:Yuan Hong and Yang Yu ; critical revision of the article for important intellectual content:Yuan Hong and Bo Chen ; study supervision: Bo Chen and Maoming Xiong. All the authors approved the final article. All authors reviewed the manuscript. Acknowledgement: Not applicable. Data Availability All data that support the findings of this study are included in this manuscript and its supplementary information files. References Rosenberg IH. Sarcopenia: origins and clinical relevance. Nutr.1997;127(5 Suppl):990S-991S. PMID: 9094910. Cruz-Jentoft AJ, Bahat G, Bauer J, et al. Sarcopenia: revised European consensus on definition and diagnosis. Age Ageing. 2019;48(4):601–7. PMID: 30817055. Chen LK, Woo J, Assantachai P et al. Asian Working Group for Sarcopenia: 2019 Consensus Update on Sarcopenia Diagnosis and Treatment. Am. Med. Dir. Assoc. 2020;21(3):300–307.e2. PMID: 30852612. Shachar SS, Williams GR, Muss HB, et al. Prognostic value of sarcopenia in adults with solid tumours: A meta-analysis and systematic review. Eur J Cancer. 2016;57:58–67. PMID: 26875329. Kamarajah SK, Bundred J, Tan BHL. Body composition assessment and sarcopenia in patients with gastric cancer: a systematic review and meta-analysis. Gastric. Cancer. 2019;22(1):10–22.PMID: 30210400. Prado CM, Baracos VE. Sarcopenia as a determinant of chemotherapy toxicity and time to tumor progression in metastatic breast cancer patients receiving capecitabine treatment. Clin Cancer Res. 2009;15(8):2920–6. PMID: 19299545. Rier HN, Jager A, Sleijfer S, et al. The Prevalence and Prognostic Value of Low Muscle Mass in Cancer Patients: A Review of the Literature. Oncologist. 2016;21(11):1396–409. PMID: 27595293. Gillies RJ, Kinahan PE, Hricak H. Radiomics: Images Are More than Pictures, They Are Data. Radiology. 2016;278(2):563–77. PMID: 26572934. Lambin P, Rios-Velazquez E, Leijenaar R, et al. Radiomics: extracting more information from medical images using advanced feature analysis. Eur J Cancer. 2012;48(4):441–6. PMID: 22445741. Martin L, Birdsell L, Macdonald N, et al. Cancer cachexia in the age of obesity: skeletal muscle depletion is a powerful prognostic factor, independent of body mass index. J Clin Oncol. 2013;31(12):1539–47. Bahat G, Turkmen BO, Aliyev S, Catikkas NM, Bakir B, Karan MA. Cut-off values of skeletal muscle index and psoas muscle index at L3 vertebra level by computerized tomography to assess low muscle mass. Clin Nutr. 2021;40(6):4360–5. Epub 2021 Jan 16. PMID: 33516603. Choi W, Kim CH, Yoo H, Yun HR, Kim DW, Kim JW. Development and validation of a reliable method for automated measurements of psoas muscle volume in CT scans using deep learning-based segmentation: a cross-sectional study. BMJ Open. 2024;14(5):e079417. 10.1136/bmjopen-2023-079417 . PMID: 38777592; PMCID: PMC11116865. Huang CB, Lin DD, Huang JQ, Hu W. Based on CT at the third lumbar spine level, the skeletal muscle index and psoas muscle index can predict osteoporosis. BMC Musculoskelet Disord. 2022;23(1):933. 10.1186/s12891-022-05887-5 . PMID: 36280811; PMCID: PMC9590212. Hu JS, Jin YP, Wu JK, Ni JG. Skeletal muscle index based on CT at the 12th thoracic spine level can predict osteoporosis and fracture risk: a propensity score-matched cohort study. Front Med (Lausanne). 2024;11:1387807. 10.3389/fmed.2024.1387807 . PMID: 38725469; PMCID: PMC11079204. Prado CM, Lieffers JR, McCargar LJ, Reiman T, Sawyer MB, Martin L, et al. Prevalence and clinical implications of sarcopenic obesity in patients with solid tumours of the respiratory and gastrointestinal tracts: a population-based study. Lancet Oncol. 2008;9(7):629–35. 10.1016/S1470-2045(08)70153-0 . Mitsiopoulos N, Baumgartner RN, Heymsfield SB, Lyons W, Gallagher D, Ross R. Cadaver validation of skeletal muscle measurement by magnetic resonance imaging and computerized tomography. J Appl Physiol (1985). 1998;85(1):115–22. 10.1152/jappl.1998.85.1.115 . Tschann P, Weigl MP, Clemens P, Szeverinski P, Attenberger C, Kowatsch M, Jäger T, Emmanuel K, Brock T, Königsrainer I. Sarcopenic Obesity Is a Risk Factor for Worse Oncological Long-Term Outcome in Locally Advanced Rectal Cancer Patients: A Retrospective Single-Center Cohort Study. Nutrients. 2023;15(11):2632. 10.3390/nu15112632 . Grotenhuis BA, Shapiro J, van Adrichem S, de Vries M, Koek M, Wijnhoven BP, van Lanschot JJB. Sarcopenia/Muscle Mass Is Not a Prognostic Factor for Short-and Long-Term Outcome After Esophagectomy for Cancer. World J Surg. 2016;40(11):2698–704. 10.1007/s00268-016-3603-1 . Prado CM, Lieffers JR, McCargar LJ, Reiman T, Sawyer MB, Martin L, et al. Prevalence and clinical implications of sarcopenic obesity in patients with solid tumours of the respiratory and gastrointestinal tracts: a population-based study. Lancet Oncol. 2008;9(7):629–35. 10.1016/S1470-2045(08)70153-0 . Marra M, Sammarco R, De Lorenzo A, Iellamo F, Siervo M, et al. Assessment of Body Composition in Health and Disease Using Bioelectrical Impedance Analysis (BIA) and Dual Energy X-Ray Absorptiometry (DXA): A Critical Overview. Contrast Media Mol Imaging. 2019;2019:3548284. 10.1155/2019/3548284 . Langius JAE, Kruizenga HM, van den Boogaardt J, et al. Comparison of the prognostic value of CT-determined sarcopenia and psoas muscle area in patients with advanced cancer. Support Care Cancer. 2018;26(5):1611–9. Martin L, Birdsell L, Macdonald N, et al. Sarcopenia defined by measured appendicular lean mass is associated with postoperative infection and delayed recovery from colorectal cancer resection surgery. Br J Cancer. 2012;107(9):931–6. Kocak B, Baessler B, Bakas S, Cuocolo R, Fedorov A, Maier-Hein L, Mercaldo N, Müller H, Orlhac F, Pinto Dos Santos D, Stanzione A, Ugga L, Zwanenburg A. CheckList for EvaluAtion of Radiomics research (CLEAR): a step-by-step reporting guideline for authors and reviewers endorsed by ESR and EuSoMII. Insights Imaging. 2023;14(1):75. 10.1186/s13244-023-01415-8 . Kocak B, Akinci D'Antonoli T, Mercaldo N, et al. METhodological RadiomICs Score (METRICS): a quality scoring tool for radiomics research endorsed by EuSoMII. Insights into Imaging. 2024;15(1):8. 10.1186/s13244-023-01572-w . PMID: 38228979; PMCID: PMC10792137. Pinto Almeida M, Fotsing G, Gijs E, Barigou M. Nutrition clinique: ce qui a changé en 2022 [Clinical nutrition: what's new in 2022]. Rev. Med. Suisse.2023;19(N°809 – 10):46–51. French. 10.53738/REVMED.2023.19.809-10.46 Cruz-Jentoft AJ, Baeyens JP, Bauer JM, et al. Sarcopenia: European consensus on definition and diagnosis: report of the European Working Group on Sarcopenia in Older People. Age Ageing. 2010;39(4):412–23. 10.1093/ageing/afq034 . Malmstrom TK, Miller DK, Simonsick EM, et al. SARC-F: a symptom score approach for the screening of sarcopenia. J Gerontol Biol Sci Med Sci. 2016;71(5):629–34. 10.1093/gerona/glv215 . Baracos VE, Martin L, Korc M, et al. Cancer-associated cachexia. Nat Rev Dis Primers. 2018;4:17105. 10.1038/nrdp.2017.105 . Muscaritoli M, Anker SD, Argilés J, et al. Consensus definition of sarcopenia, cachexia and pre-cachexia: joint document elaborated by Special Interest Groups (SIG) cachexia-anorexia in chronic wasting diseases and nutrition in geriatrics. Clin Nutr. 2010;29(2):154–9. 10.1016/j.clnu.2009.12.004 . Amini B, Boyle SP, Boutin RD, et al. Approaches to assessment of muscle mass and myosteatosis on computed tomography: a systematic review. Gerontol Biol Sci Med Sci. 2019;74(7):1079–86. 10.1093/gerona/glz034 . Jang MK, Park S, Raszewski R, et al. Prevalence and clinical implications of sarcopenia in breast cancer: a systematic review and meta-analysis. Support Care Cancer. 2024;32:328. https://doi.org/10.1007/s00520-024-08532-0 . Kasahara K, Kono T, Sato Y, Ueno M, So H, Fuse Y, Shinden S, Ozawa H. Sarcopenia accompanied by systemic inflammation can predict clinical outcomes in patients with head and neck cancer undergoing curative therapy. Front Oncol. 2024;14:1378762. 10.3389/fonc.2024.1378762 . PMID: 38549928; PMCID: PMC10973154. Prado CM, Heymsfield SB. Lean tissue imaging: a new era for nutritional assessment and intervention. JPEN J Parenter Enter Nutr. 2014;38(8):940–53. 10.1177/0148607114550189 . He J, Huang Y, Huang N, Jiang J. Prevalence and predictive value of sarcopenia in surgically treated cholangiocarcinoma: a comprehensive review and meta-analysis. Front Oncol 14:1363843. 10.3389/fonc.2024.1363843 Grotenhuis BA, Shapiro J, van Adrichem S, et al. Sarcopenia/muscle mass is not a prognostic factor for short-and long-term outcome after esophagectomy for cancer. World J Surg. 2016;40(7):2698–704. 10.1007/s00268-016-3603-1 . Chen RH, Chang K, West RB. Clinical Applications of Machine Learning in Oncology. Nat Reviews Clin Oncol. 2018;15(9):568–86. 10.1038/s41571-018-0004-z . Zhang B, Tian J, Dong D, et al. Radiomics Features of Multiparametric MRI as Novel Prognostic Factors in Advanced Nasopharyngeal Carcinoma. Clin Cancer Res. 2017;23(15):4259–69. 10.1158/1078-0432.CCR-16-2910 . Tables Tables 1 to 11 are available in the Supplementary Files section. Additional Declarations No competing interests reported. Supplementary Files Tablefiles.docx ClinicalInclusionandExclusionCriteria.docx Cite Share Download PDF Status: Under Revision Version 1 posted Editorial decision: Revision requested 11 Nov, 2024 Editor assigned by journal 06 Nov, 2024 Submission checks completed at journal 06 Nov, 2024 First submitted to journal 28 Oct, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5350805","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":376764507,"identity":"a94babf1-0439-451f-b48c-53e3052da808","order_by":0,"name":"Yuan Hong","email":"","orcid":"","institution":"The First Affiliated Hospital of Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yuan","middleName":"","lastName":"Hong","suffix":""},{"id":376764508,"identity":"4ed53fd2-6ac7-4e06-8b8a-4081ee666ec2","order_by":1,"name":"Yifan Li","email":"","orcid":"","institution":"The First Affiliated Hospital of Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yifan","middleName":"","lastName":"Li","suffix":""},{"id":376764509,"identity":"4f0527a3-5ca8-4894-8468-d38c1654fd9b","order_by":2,"name":"Peng Zhang","email":"","orcid":"","institution":"The First Affiliated Hospital of Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Peng","middleName":"","lastName":"Zhang","suffix":""},{"id":376764510,"identity":"c9ab7ccd-bb12-49cd-9f54-038fcecb1b9e","order_by":3,"name":"Haosong Chen","email":"","orcid":"","institution":"The First Affiliated Hospital of Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Haosong","middleName":"","lastName":"Chen","suffix":""},{"id":376764511,"identity":"258fcae0-d747-490b-bcbf-ac0b52841eba","order_by":4,"name":"Yixian Chen","email":"","orcid":"","institution":"The First Affiliated Hospital of Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yixian","middleName":"","lastName":"Chen","suffix":""},{"id":376764512,"identity":"e33ffb8c-cf97-4348-bdc2-acb373dbd81a","order_by":5,"name":"Yang Yu","email":"","orcid":"","institution":"The First Affiliated Hospital of Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yang","middleName":"","lastName":"Yu","suffix":""},{"id":376764513,"identity":"7019611d-225f-48d5-8626-6e6cdfaf3dfc","order_by":6,"name":"Zimo Zhang","email":"","orcid":"","institution":"Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zimo","middleName":"","lastName":"Zhang","suffix":""},{"id":376764514,"identity":"d621a6f8-71d7-49d0-ab2d-4cede18adeb5","order_by":7,"name":"Kang Cheng","email":"","orcid":"","institution":"Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Kang","middleName":"","lastName":"Cheng","suffix":""},{"id":376764515,"identity":"4bf1136c-f8b2-423f-af9b-80028b7c70b6","order_by":8,"name":"Maoming Xiong","email":"","orcid":"","institution":"The First Affiliated Hospital of Anhui Medical University","correspondingAuthor":false,"prefix":"","firstName":"Maoming","middleName":"","lastName":"Xiong","suffix":""},{"id":376764516,"identity":"907ca347-42f9-4613-ba13-d1540ce3ed16","order_by":9,"name":"Bo Chen","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAy0lEQVRIiWNgGAWjYDACCQYGZiApx8/ffODAhx/Ea7EwlpxxLPHgzB7itVQkbjiQY3yYg40IHfyzm489LqiQSJzZcObDYQYeBnl+sQMELLlzLN14xhkJ437m3g2HCywYDGfOTsCvxUAix0yat01CdmbD2Q2HZ/AwJBjcJqgl/xtICyPQLw8O87ARpSWHDaRFEaiFgTgtEjfSzKRBfgEGsgEwkCUI+4V/RvIz6YKKOlBUPv7w4YeNPL80AS0YtpKmfBSMglEwCkYBdgAAxcZFfo4i1b8AAAAASUVORK5CYII=","orcid":"","institution":"The First Affiliated Hospital of Anhui Medical University","correspondingAuthor":true,"prefix":"","firstName":"Bo","middleName":"","lastName":"Chen","suffix":""}],"badges":[],"createdAt":"2024-10-29 03:53:16","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5350805/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5350805/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":70922443,"identity":"14ae6954-5b29-4c04-981a-bb3b8ca13791","added_by":"auto","created_at":"2024-12-09 08:49:32","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":3703540,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eOutline The Bilateral Psoas Major Muscles and Calculate The Area\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAxial CT at the L3 level and segmentation of the bilateral psoas major muscle.This manual segmentation of skeletal muscles was performed using 3D-Slicer software(version 5.6.1,Boston,the USA).The square of the bilateral psoas major muscle are annotated in the figure. CT: computed tomography; L: lumbar.\u003c/p\u003e","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/2274713af510472db8e8aedf.png"},{"id":70924613,"identity":"1499d9c6-6998-4608-b988-155113f66c15","added_by":"auto","created_at":"2024-12-09 09:05:32","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":4858916,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRadiomics Workflow Diagram\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/ed0548fae514e525e3ce14a8.png"},{"id":70922441,"identity":"e99faa9c-4bbd-4ba6-a7db-a25606a9d232","added_by":"auto","created_at":"2024-12-09 08:49:32","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2322756,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e3D Muscle Sketch Flow Chart\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/6d509a11e360945e7db15b30.png"},{"id":70923967,"identity":"11e1960f-710e-4029-8833-e05c52ac26e8","added_by":"auto","created_at":"2024-12-09 08:57:32","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":117741,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRaw Feature Statistics (Unfiltered)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe outcomes of feature extraction encompass:glszm, gray-level size zone matrix; glrlm, gray level run length matrix ; gldm, gray level dependence matrix; glcm, gray level co-occurrence matrix; firstorder, first-order statistics; ngtdm, neighbouring gray tone difference matrix.\u003c/p\u003e","description":"","filename":"Fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/ea1d9e29f45e3dddb79cac3c.png"},{"id":70924611,"identity":"f6a5cc99-49d9-40a0-991e-1af9cd838432","added_by":"auto","created_at":"2024-12-09 09:05:32","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1213789,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eKM Analysis Results of Sarcopenia and Non-Sarcopenia Groups\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe 5-year Kaplan-Meier (K-M) survival curves, along with their 95% confidence intervals, were compared between the sarcopenia group and the non-sarcopenia group.\u003c/p\u003e","description":"","filename":"Fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/7ec5c47979f0b80c0efe800f.png"},{"id":70922446,"identity":"4b77d833-71c2-4620-a013-3c41f62f3cfe","added_by":"auto","created_at":"2024-12-09 08:49:32","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1502077,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC for KNN and NaiveBayes models\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Fig6.png","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/7360c4c183e16259011e224d.png"},{"id":70922449,"identity":"59b40d5b-ded0-4c56-ae2e-3c5454af3578","added_by":"auto","created_at":"2024-12-09 08:49:32","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":3023122,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC of RandomForest, ExtraTrees, GradientBoosting and AdaBoost models\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Fig7.png","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/2727e8e46344af037d7afe5a.png"},{"id":70923973,"identity":"92d71435-1403-4513-8aea-bdf33efc1a4d","added_by":"auto","created_at":"2024-12-09 08:57:32","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":443051,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC for SVM and XGBoost models\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Fig8.png","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/75cea69362ef392b39ea50b8.png"},{"id":70923970,"identity":"2c3a78d6-b6cd-4441-8457-2eedcb2424b6","added_by":"auto","created_at":"2024-12-09 08:57:32","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":1040950,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC of LR, LightGBM and MLP models\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Fig9.png","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/ac7bad2bc64a53b811a6a14a.png"},{"id":70922453,"identity":"c86bb89b-98d7-40a6-a8b3-40e64819f396","added_by":"auto","created_at":"2024-12-09 08:49:32","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":375028,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC for LR and Confusion LR models\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Fig10.png","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/04ecf89b5ab816139ddebd47.png"},{"id":70926662,"identity":"b96dc322-cac1-454b-82cd-a71d6fdf637a","added_by":"auto","created_at":"2024-12-09 09:13:32","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":2568986,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDCA for LR and Confusion LR models\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Fig11.png","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/de20a027413d4d9948f84420.png"},{"id":70926675,"identity":"97fa629b-61d6-4773-be33-dad3b7ecf7ca","added_by":"auto","created_at":"2024-12-09 09:13:53","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":24587102,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/de4dfa44-b6ae-420f-ac69-853e1d4b9bdf.pdf"},{"id":70923968,"identity":"e17b6227-383d-42c1-9410-3c59263fd66e","added_by":"auto","created_at":"2024-12-09 08:57:32","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":157530,"visible":true,"origin":"","legend":"","description":"","filename":"Tablefiles.docx","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/a259dc4dad9f0c317e96045e.docx"},{"id":70924612,"identity":"5ea23f93-7a78-4f62-9c6d-6dfb864bf866","added_by":"auto","created_at":"2024-12-09 09:05:32","extension":"docx","order_by":14,"title":"","display":"","copyAsset":false,"role":"supplement","size":13066,"visible":true,"origin":"","legend":"","description":"","filename":"ClinicalInclusionandExclusionCriteria.docx","url":"https://assets-eu.researchsquare.com/files/rs-5350805/v1/15e1eca94b17e8bf676d6afe.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Predicting 5-Year Survival in Gastric Cancer Patients Using Iliopsoas Muscle CT Radiomics and Machine Learning Techniques","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eSarcopenia, initially proposed by Irwin Rosenberg in the 1980s[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e], refers to age-related muscle loss. Over time, its definition has evolved to include degenerative changes in skeletal muscle mass, strength, and function. In 2022, the International Sarcopenia Working Group provided a detailed definition, describing it as a progressive reduction in muscle mass, strength, or physiological function associated with aging.[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e] The European and Asian Working Groups played crucial roles in standardizing diagnostic criteria, which now encompass assessments of muscle strength, mass, and physical performance.[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e] A meta-analysis of 38 studies (7,843 patients) revealed that low skeletal muscle index (SMI) at diagnosis predicts poor survival, especially in gastric and esophageal cancer patients.[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e] Gastric cancer patients have a higher sarcopenia incidence, linked to age, gender, nutritional status, and disease stage.[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e] Gastric cancer-induced metabolic changes, including imbalances in carbohydrate, protein, and fat metabolism, contribute to sarcopenia.[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]Given that the diagnosis of sarcopenia primarily relies on the cross-sectional data of the psoas major muscle, and previous research on sarcopenia has commonly focused on the iliac psoas muscle as the target muscle, this study also follows this research path. The iliac psoas muscle was considered the core research subject, and precise delineation was performed using two-dimensional sections and three-dimensional models.[\u003cspan additionalcitationids=\"CR2 CR3 CR4 CR5\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/p\u003e \u003cp\u003eDespite the link between sarcopenia and gastric cancer, research on using sarcopenia to reveal clinical information remains limited. Regional variations in diagnostic standards, traditional muscle mass assessment limitations, and insufficient clinical awareness contribute to this gap.[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/p\u003e \u003cp\u003eRadiomics, an emerging technology, extracts multidimensional features from medical images (such as CT and MRI) using automated algorithms. It supports precise disease diagnosis, classification, staging, and prognosis prediction. In sarcopenia research, radiomics aims for intelligent, accurate diagnosis and direct application in assessing patient survival and prognosis.[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e] However, overall research in this field remains limited, and the application of AI technology requires further expansion.[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]\u003c/p\u003e \u003cp\u003eOur objective is to integrate AI algorithms for accurate prediction of 5-year survival in gastric cancer patients. Leveraging AI technology, we aim to enhance personalized treatment and prognosis management.\u003c/p\u003e"},{"header":"2. Material and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1. Clinical Data Collection\u003c/h2\u003e \u003cp\u003e This study, approved by the Ethics Committee, retrospectively analyzed clinical data from 100 gastric cancer patients diagnosed between December 2018 and June 2019. The detailed inclusion and exclusion criteria for patients are shown in the supplementary materials. After radical gastrectomy, continuous follow-up was conducted. Of the initial sample, 12 patients declined participation, and 17 were lost to follow-up. However, five-year survival data for the remaining 71 patients were accurately recorded. To validate the performance of the model and mitigate the risk of overfitting, our center, upon obtaining authorization, utilized preoperative CT and clinical data from 34 patients diagnosed with gastric cancer who underwent major gastric surgery at Hanshan County People\u0026rsquo;s Hospital between 2017 and 2018 as an external validation set. The inclusion criteria for this validation set were consistent with our center\u0026rsquo;s internal standards to ensure data comparability. All patients completed a 5-year follow-up, during which survival data were collected.\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eThe CT dataset for this study was obtained from routine preoperative scans using a 256-channel scanner (Philips Brilliance i CT). Image acquisition utilized a 512 \u0026times; 512 matrix, maintaining in-plane resolution between 0.62 \u0026times; 0.62 mm and 0.86 \u0026times; 0.86 mm. To enhance image visualization, 3D Slicer software adjusted the CT image window width (WW) to 149 and window level (WL) to 40, with a threshold range of -35 to 115.[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2. Assessment of Sarcopenia\u003c/h2\u003e \u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eIn the current research on sarcopenia diagnosis, most studies prefer using non-contrast computed tomography (CT) with specific abdominal window settings for image analysis. This method effectively distinguishes muscle tissue from surrounding structures, providing detailed muscle imaging. It significantly enhances the accuracy of muscle mass and quality assessment. Our study also employs this standard CT imaging technique for detailed muscle morphology analysis.[\u003cspan additionalcitationids=\"CR12 CR13\" citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eThe gold standard for diagnosing sarcopenia involves CT imaging, specifically measuring skeletal muscle area at the L3 vertebral level and calculating the SMI.[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]A newly introduced method uses bioelectrical impedance analysis (BIA) to assess muscle content.[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e] The psoas muscle index (TPI) is also valuable, standardized by measuring psoas muscle area at L3.[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]TPI assessments are more practical than BIA, which can be influenced by body water content. [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]Our study adheres to TPI standards, considering clinical examination results and dual criteria (6-meter gait speed and grip strength) for high-risk sarcopenia patients. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e outlines the psoas muscle delineation process at the bilateral L3 plane, ensuring reproducibility and study result accuracy.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eTo diagnose sarcopenia, we adhere to international consensus standards for cancer patients.[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e] Sarcopenia is diagnosed when TPI falls below 385 mm\u0026sup2;/m\u0026sup2; for females or 545 mm\u0026sup2;/m\u0026sup2; for males.[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e] We used the lasso tool in 3D Slicer 5.4.0 software to outline 2D images, with collaborative input from experienced radiologists and final review by a senior physician to ensure measurement accuracy and reliability.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3. Clinical Data Statistical Analysis\u003c/h2\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eIn this study, we diagnosed sarcopenia using TPI at the L3 level, dividing patients into sarcopenia and non-sarcopenia groups. We collected 11 clinical indicators, including height, weight, and TNM staging, for the training set and conducted statistical analysis. The external validation set, limited by data authorization, only includes basic pathological and clinical information, so no in-depth analysis was performed. Correlation analysis revealed no significant association between clinical factors and sarcopenia diagnosis in training set. To explore survival factors, we used Kaplan-Meier survival analysis and univariate COX regression. Multivariate COX regression identified significant survival factors. Combining these with radiomics features, we trained and tested machine learning models, ultimately selecting an accurate 5-year survival prediction model for gastric cancer patients. To ensure the validity of the analysis results, the research team conducted a post-hoc power test on patient survival time after completing the statistical analysis.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4. Radiomics Workflow and Muscle Segmentation\u003c/h2\u003e \u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eThe overall radiomics workflow is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. In the study, abdominal plain CT images were used. Researchers manually delineated the bilateral iliopsoas muscles using 3dsclier software with standard abdominal window settings. The specific 3D delineation process is illustrated in Fig.\u0026nbsp;3. To ensure accuracy and minimize human error, two radiologists independently performed the manual delineation, which was then reviewed and fine-tuned by a senior radiologist. We used the pyradiomics software package in Python to extract radiomic features from the 3D iliopsoas muscle images of the patients. Radiomic features were obtained using fixed bin number discretization (bin number\u0026thinsp;=\u0026thinsp;5) and voxel resampling (X, Y, Z\u0026thinsp;=\u0026thinsp;3 \u0026times; 3 \u0026times; 3 mm\u0026sup3;).After feature extraction, we normalized features for subsequent data processing. We assessed feature correlation using the T-test and Spearman coefficient. Highly correlated features (coefficient\u0026thinsp;\u0026gt;\u0026thinsp;0.9) were reduced to simplify model complexity. A Lasso regression model screened significant prediction-contributing features. The feature engineering process is detailed in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, and feature statistics are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e. Machine learning algorithms excel at handling large-scale, high-dimensional clinical information. They uncover complex patterns to achieve precise predictions, surpassing traditional statistical methods and enhancing disease diagnosis accuracy and efficiency. We innovatively apply these algorithms to prognosis modeling for sarcopenia patients, evaluating predictive performance across 11 selected algorithms to identify the optimal model for analyzing patient imaging features. Key features were input into 11 machine learning algorithms (SVM:support vector machine, KNN:k-nearest neighbors, RandomForest, ExtraTrees:extremely randomized trees, XGBoost, LightGBM:light gradient boosting machine, NaiveBayes: naive bayes classifier, AdaBoost:adaptive boosting, GradientBoosting, LR:logistic regression, MLP:multi-layer perceptron). We randomly split the patient dataset (8:2) for training and testing. After completing the model training phase, we evaluated the extrapolation performance of the constructed model and monitored the risk of overfitting using external validation set data. Based on the performance results on the validation set, we systematically screened the models and ultimately identified the best-performing model.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eWe evaluated model performance using ROC curves on training and test sets, comparing predictive accuracy via AUC. In constructing the 5-year survival prediction model, we analyzed the impact of clinical and radiomic features, as well as their fusion, across different algorithm frameworks. We utilize the CLEAR[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e] form and METRICS tool[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e] to provide a comprehensive assessment of the quality of imaging work, please see the attached Supplemental Checklist 1 and Supplemental Checklist 2 for specific assessment details.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e, Radiomics Feature Extraction Settings\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSetting\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDetermination\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBin Method\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFBN\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBin Amount\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMethod\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSitkNearestNeighbor\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResample Filter\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResample Spacing X\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3mm\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResample Spacing Y\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3mm\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResample Spacing Z\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3mm\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"2\"\u003eFBN,fixed bin number\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5. Statistics\u003c/h2\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eIn the clinical data statistics phase, we securely saved patient information in anonymized Excel format. Statistical analysis used SPSS 27.0 for categorical variables (chi-square test) and continuous variables (t-test or Mann-Whitney U test). Survival analysis employed Kaplan-Meier (KM) and log-rank tests. Radiomics operations were implemented in Python (scikit-learn, pandas) within Jupyter Notebook. High-quality images were plotted using matplotlib. Radiomic features from imaging data were securely stored in Excel for subsequent analysis.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\n \u003ch2\u003e3.1. Diagnosis of Sarcopenia\u003c/h2\u003e\n \u003cp\u003eFollowing Pelten et al.\u0026apos;s[\u003cspan class=\"CitationRef\"\u003e25\u003c/span\u003e] criteria, we accurately located L3 vertebral plane images using 3D Slicer software. The lasso tool meticulously outlined bilateral psoas major muscles. We calculated TPI by dividing total muscle area by the square of patient height. Patients with TPI below 385 mm\u0026sup2;/m\u0026sup2; (females) or 545 mm\u0026sup2;/m\u0026sup2; (males) were diagnosed with sarcopenia. As shown in Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e, our training set included 44 sarcopenia patients (37 males, 7 females) and 27 non-sarcopenia patients (22 males, 5 females).The validation set (shown in Table \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e) included 11 sarcopenia patients (9 males, 2 females) and 23 non-sarcopenia patients (20 males, 3 females). Notably, the sarcopenia group had significantly lower psoas major index than the non-sarcopenia group for both genders.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\n \u003ch2\u003e3.2. Clinical Baseline Variability Analysis\u003c/h2\u003e\n \u003cp\u003eWe distinguished gastric cancer patients in training set into sarcopenia and non-sarcopenia groups, organizing 11 clinical data points (Table \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e). The clinical baseline data of the validation set are presented in the Table \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e. It should be noted that due to the limitations of the clinical data in the validation set, this dataset was not used for in-depth differential analysis. Statistical methods varied by data type: chi-square tests (with continuity correction, Pearson, or Fisher\u0026rsquo;s exact) for binary variables (gender, surgical method, M stage), RC contingency table chi-square for multi-category variables (T stage, N stage), t-test for height (continuous), and Mann-Whitney U test for age, weight, BMI, hospital stay, and surgery time (continuous).\u003c/p\u003e\n \u003cp\u003eStatistical analysis revealed that only surgery time significantly differed between sarcopenia and non-sarcopenia groups. Sarcopenia patients had longer surgery times. Other clinical variables showed no significant differences. This suggests that sarcopenia may complicate gastric cancer surgery, warranting further research into underlying mechanisms and impacts.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\n \u003ch2\u003e3.3. Clinical Survival Analysis\u003c/h2\u003e\n \u003cp\u003eWe analyzed 5-year survival data using Kaplan-Meier (KM) survival analysis for sarcopenia and non-sarcopenia groups (Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e). No significant difference in 5-year survival rates was observed. Correlation tests on clinical factors (Supplementary Table\u0026nbsp;1) revealed no significant association with sarcopenia.\u003c/p\u003e\n \u003cp\u003eWe performed univariate Cox regression analysis to assess the impact of various factors on patient survival. Age, weight, BMI, T stage, N stage, and M stage all significantly affected survival. Collinearity diagnostics confirmed independence among these factors, with variance inflation factors (VIF) below 5 (Tables \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003e and\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003e).\u003c/p\u003e\n \u003cp\u003eWe incorporated six significant factors (age, weight, BMI, T stage, N stage, and M stage) into a multivariate Cox regression model. The results highlighted age and tumor M stage as independent and significant factors affecting the 5-year survival rate of gastric cancer patients (Table \u003cspan class=\"InternalRef\"\u003e8\u003c/span\u003e).A post-hoc power test was conducted on patient survival time, with a diagnostic efficiency of 0.834, confirming the validity of the statistical analysis. Detailed results are provided in Table \u003cspan class=\"InternalRef\"\u003e9\u003c/span\u003e.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\n \u003ch2\u003e3.4. Predictive Modelling of 5-year Survival\u003c/h2\u003e\n \u003cp\u003eWe transformed patients\u0026rsquo; five-year survival status into binary data: \u0026lsquo;1\u0026rsquo; for patients surviving more than five years and \u0026lsquo;0\u0026rsquo; for those who passed away within the follow-up period. We selected radiological features and clinical variables as input, using 11 machine learning models. Comprehensive evaluation occurred via five-fold cross-validation.The classification performance and generalizability of each model is demonstrated in the external validation set.\u003c/p\u003e\n \u003cp\u003eIn this paper, we present the performance results of 11 models on both the training and test sets, as detailed in the Table \u003cspan class=\"InternalRef\"\u003e10\u003c/span\u003e below. In terms of model accuracy, the external validation accuracy of both the Naive Bayes and K-Nearest Neighbors (KNN) models is below 0.5, indicating that the predictive performance of these two models is unacceptable. The ROC images of the two models are shown in Fig. \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003e. Upon analysis, the KNN model exhibited excessive prediction bias and an abnormally high AUC, indicating poor generalization ability and insufficient predictive accuracy. Therefore, the research team decided to discard this model and explore other superior models.\u003c/p\u003e\n \u003cp\u003eFurthermore, by analyzing the ROC curves of each model (as shown in the Fig. \u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003e), we found that the AUC values of AdaBoost, Extra Trees, Gradient Boosting, Random Forest, Support Vector Machine (SVM), and XGBoost on the test set all exceeded 0.95, demonstrating extremely high predictive capabilities. However, considering the results on the validation set, the AUC values of the AdaBoost and Gradient Boosting models on the validation set were close to 0.5, indicating significant overfitting to the training set. Additionally, the performance of the Random Forest and Extra Trees models on the external validation set was also poor, raising doubts about their classification performance.\u003c/p\u003e\n \u003cp\u003eFor the SVM and XGBoost models, their accuracy and AUC values on the validation set were both stable around 0.7. The ROC images of the two models are shown in Fig. \u003cspan class=\"InternalRef\"\u003e8\u003c/span\u003e.However, given the excessively high AUC values on the training set, these two models still present a significant risk of overfitting. Although they show potential for achieving high-quality classification models with larger training data volumes, due to the data limitations of this study, we currently do not recommend these two models as the optimal choices.\u003c/p\u003e\n \u003cp\u003eAmong the remaining three models, Logistic Regression (LR), LightGBM, and Multilayer Perceptron (MLP) exhibited stable AUC and accuracy on both the training and test sets, with relatively low risk of overfitting.The ROC images for the three sets of models are shown in Fig. \u003cspan class=\"InternalRef\"\u003e9\u003c/span\u003e. By comprehensively comparing the classification performance of these three models, we found that the LR model maintained an accuracy and AUC value of around 0.7 on both the training and test sets, demonstrating good classification performance. Therefore, we consider the LR model to have performed the best in this study.\u003c/p\u003e\n \u003cp\u003eUltimately, we systematically integrated the collected clinical independent factors. Subsequently, to explore potential ways to enhance the model\u0026rsquo;s predictive ability, we combined these clinical features with imaging features. A comparison of the ROC curves of the two groups of models is shown in Fig. \u003cspan class=\"InternalRef\"\u003e10\u003c/span\u003e and the model accuracy is detailed in Table \u003cspan class=\"InternalRef\"\u003e11\u003c/span\u003e.However, analysis of the fused model revealed an interesting phenomenon: although the classification accuracy of this model improved on the internal test set, its performance on the external validation set significantly declined. Based on this observation, we cautiously concluded that while the introduction of clinical features unique to our center enhanced the model\u0026rsquo;s internal test classification performance, its generalization ability correspondingly weakened. As shown in Fig. \u003cspan class=\"InternalRef\"\u003e11\u003c/span\u003e, analysis of the DCA curves for the two groups of models shows that incorporating clinical features enhances net benefits in the training and internal test sets. However, the performance in the external validation set is suboptimal, slightly lower than that of the non-integrated model. This indicates that the performance improvement from integrating clinical features is not consistently demonstrated in external validation.\u003c/p\u003e\n \u003cp\u003eTherefore, in the specific context of this study, we evaluated that the LR model without clinical feature fusion exhibited the best performance, achieving approximately 70% accuracy on new datasets, providing valuable reference for our subsequent research.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eSarcopenia, characterized by the loss of skeletal muscle mass and function, is a critical factor influencing clinical outcomes in oncology patients. The diagnosis of sarcopenia has evolved with several mainstream methods being utilized. Commonly used screening tools include the five-item questionnaires SARC-F and SARC-CalF.[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e] Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) are considered gold standards due to their accuracy in quantifying muscle mass[\u003cspan additionalcitationids=\"CR29\" citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e], particularly the Skeletal Muscle Index (SMI) at the lumbar vertebra level[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. Dual-energy X-ray Absorptiometry (DXA) is another widely used method, offering a balance between accuracy and accessibility[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Bioelectrical Impedance Analysis (BIA) provides a non-invasive and cost-effective alternative, though it is less precise compared to imaging techniques[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. Early identification and management of sarcopenia are crucial, as it is associated with increased chemotherapy toxicity, postoperative complications, and reduced survival rates[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Therefore, integrating these diagnostic methods into routine clinical practice can significantly enhance patient care and treatment outcomes. Prognostic modeling in sarcopenia among oncology patients has advanced significantly in recent years. Current mainstream models incorporate both sarcopenia and systemic inflammation markers, such as the neutrophil/lymphocyte ratio (NLR) and lymphocyte/monocyte ratio (LMR), to predict clinical outcomes. Studies have shown that combining sarcopenia with low LMR is a sensitive prognostic factor for overall survival (OS) and progression-free survival (PFS) in head and neck cancer patients[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. Additionally, sarcopenia has been identified as a negative predictive factor for survival outcomes and postoperative complications in cholangiocarcinoma[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. These models underscore the importance of integrating muscle mass assessments with inflammatory markers to enhance prognostic accuracy in oncology patients[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e].\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eIn our clinical cohort, patients with sarcopenia exhibited significantly longer surgery times compared to those without sarcopenia, even when undergoing the same surgical method. This suggests a potential link between sarcopenia and malignant tumors. Although correlation analysis of clinical factors was conducted, no significant association with sarcopenia was found[\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e], possibly due to the limited sample size and limitations of traditional statistics. Additionally, in-depth analysis of 5-year survival data revealed the impact of patient age and tumor M stage on prognosis. An independent prediction model highlighted the key role of clinical factors in assessing long-term survival. However, after applying clinical modality fusion, we observed an unexpected decline in the model\u0026rsquo;s extrapolation performance. This phenomenon indirectly suggests that the independent influencing factors selected using our center\u0026rsquo;s data may have suffered from a certain degree of training set overfitting. Therefore, this research team does not recommend directly applying the selected clinical independent factors to new datasets. Regarding the actual clinical significance of the two selected clinical factors, we plan to conduct more in-depth clinical data collection and analysis to gain a more comprehensive understanding.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThis study applied muscle radiomics features from gastric cancer patients to predict medium- and long-term survival. The model demonstrated good classification performance in the classification tasks and exhibited stable extrapolation capability, thereby strongly proving the great potential of artificial intelligence technology in clinical applications. [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]Accurate prediction of overall survival time based on muscle mass is crucial for personalized treatment planning.\u003c/p\u003e \u003cp\u003eThis study acknowledges limitations due to a small sample size and single-center design, potentially affecting model accuracy. Efforts will focus on expanding the clinical cohort and collecting high-quality data. While current models rely on traditional machine learning omics algorithms, the research group aims to enhance performance by exploring advanced radiomics techniques, including Convolutional Neural Networks (CNNs). Expectations are high that AI technology will contribute to sarcopenia research in oncology.\u003c/p\u003e"},{"header":"5. Conclusions","content":"\u003cp\u003eIn this study, we developed a machine learning model based on preoperative CT imaging data of gastric cancer patients to predict their five-year survival rate. The model can achieve about 70% accuracy. Additionally, we explored the necessity and rationale of incorporating clinical independent factors into this predictive model. The results indicated a significant correlation between muscle imaging features and overall patient survival, highlighting the importance of sarcopenia in the clinical management of gastric cancer patients. This study not only provides new tools and methods for prognostic evaluation of gastric cancer patients but also emphasizes the significance of artificial intelligence technology in enhancing the quality of gastric cancer treatment decisions.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eFunding:\u003c/h2\u003e \u003cp\u003eThis study was supported by a grant from the Graduate Student Research and Practice Innovation Program of Anhui Medical University (YJS20230080).\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eCompeting interests:\u003c/strong\u003e \u003cp\u003eThe authors declare no potential conflict of interest in the research, writing, and publication of this article.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eConcept and design:Yuan Hong and Bo Chen; data collection and analysis:Yifan Li , Peng Zhang , Haosong Chen , Yixian Cheng , Zimo Zhang and Kang Cheng; drafting of the article:Yuan Hong and Yang Yu ; critical revision of the article for important intellectual content:Yuan Hong and Bo Chen ; study supervision: Bo Chen and Maoming Xiong. All the authors approved the final article. All authors reviewed the manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement:\u003c/h2\u003e \u003cp\u003eNot applicable.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eAll data that support the findings of this study are included in this manuscript and its supplementary information files.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eRosenberg IH. Sarcopenia: origins and clinical relevance. Nutr.1997;127(5 Suppl):990S-991S. PMID: 9094910.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCruz-Jentoft AJ, Bahat G, Bauer J, et al. Sarcopenia: revised European consensus on definition and diagnosis. Age Ageing. 2019;48(4):601\u0026ndash;7. PMID: 30817055.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen LK, Woo J, Assantachai P et al. Asian Working Group for Sarcopenia: 2019 Consensus Update on Sarcopenia Diagnosis and Treatment. Am. Med. Dir. Assoc. 2020;21(3):300\u0026ndash;307.e2. PMID: 30852612.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShachar SS, Williams GR, Muss HB, et al. Prognostic value of sarcopenia in adults with solid tumours: A meta-analysis and systematic review. Eur J Cancer. 2016;57:58\u0026ndash;67. PMID: 26875329.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKamarajah SK, Bundred J, Tan BHL. Body composition assessment and sarcopenia in patients with gastric cancer: a systematic review and meta-analysis. Gastric. Cancer. 2019;22(1):10\u0026ndash;22.PMID: 30210400.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePrado CM, Baracos VE. Sarcopenia as a determinant of chemotherapy toxicity and time to tumor progression in metastatic breast cancer patients receiving capecitabine treatment. Clin Cancer Res. 2009;15(8):2920\u0026ndash;6. PMID: 19299545.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRier HN, Jager A, Sleijfer S, et al. The Prevalence and Prognostic Value of Low Muscle Mass in Cancer Patients: A Review of the Literature. Oncologist. 2016;21(11):1396\u0026ndash;409. PMID: 27595293.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGillies RJ, Kinahan PE, Hricak H. Radiomics: Images Are More than Pictures, They Are Data. Radiology. 2016;278(2):563\u0026ndash;77. PMID: 26572934.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLambin P, Rios-Velazquez E, Leijenaar R, et al. Radiomics: extracting more information from medical images using advanced feature analysis. Eur J Cancer. 2012;48(4):441\u0026ndash;6. PMID: 22445741.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMartin L, Birdsell L, Macdonald N, et al. Cancer cachexia in the age of obesity: skeletal muscle depletion is a powerful prognostic factor, independent of body mass index. J Clin Oncol. 2013;31(12):1539\u0026ndash;47.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBahat G, Turkmen BO, Aliyev S, Catikkas NM, Bakir B, Karan MA. Cut-off values of skeletal muscle index and psoas muscle index at L3 vertebra level by computerized tomography to assess low muscle mass. Clin Nutr. 2021;40(6):4360\u0026ndash;5. Epub 2021 Jan 16. PMID: 33516603.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChoi W, Kim CH, Yoo H, Yun HR, Kim DW, Kim JW. Development and validation of a reliable method for automated measurements of psoas muscle volume in CT scans using deep learning-based segmentation: a cross-sectional study. BMJ Open. 2024;14(5):e079417. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmjopen-2023-079417\u003c/span\u003e\u003cspan address=\"10.1136/bmjopen-2023-079417\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 38777592; PMCID: PMC11116865.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang CB, Lin DD, Huang JQ, Hu W. Based on CT at the third lumbar spine level, the skeletal muscle index and psoas muscle index can predict osteoporosis. BMC Musculoskelet Disord. 2022;23(1):933. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12891-022-05887-5\u003c/span\u003e\u003cspan address=\"10.1186/s12891-022-05887-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 36280811; PMCID: PMC9590212.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHu JS, Jin YP, Wu JK, Ni JG. Skeletal muscle index based on CT at the 12th thoracic spine level can predict osteoporosis and fracture risk: a propensity score-matched cohort study. Front Med (Lausanne). 2024;11:1387807. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fmed.2024.1387807\u003c/span\u003e\u003cspan address=\"10.3389/fmed.2024.1387807\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 38725469; PMCID: PMC11079204.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePrado CM, Lieffers JR, McCargar LJ, Reiman T, Sawyer MB, Martin L, et al. Prevalence and clinical implications of sarcopenic obesity in patients with solid tumours of the respiratory and gastrointestinal tracts: a population-based study. Lancet Oncol. 2008;9(7):629\u0026ndash;35. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S1470-2045(08)70153-0\u003c/span\u003e\u003cspan address=\"10.1016/S1470-2045(08)70153-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMitsiopoulos N, Baumgartner RN, Heymsfield SB, Lyons W, Gallagher D, Ross R. Cadaver validation of skeletal muscle measurement by magnetic resonance imaging and computerized tomography. J Appl Physiol (1985). 1998;85(1):115\u0026ndash;22. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1152/jappl.1998.85.1.115\u003c/span\u003e\u003cspan address=\"10.1152/jappl.1998.85.1.115\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTschann P, Weigl MP, Clemens P, Szeverinski P, Attenberger C, Kowatsch M, J\u0026auml;ger T, Emmanuel K, Brock T, K\u0026ouml;nigsrainer I. Sarcopenic Obesity Is a Risk Factor for Worse Oncological Long-Term Outcome in Locally Advanced Rectal Cancer Patients: A Retrospective Single-Center Cohort Study. Nutrients. 2023;15(11):2632. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/nu15112632\u003c/span\u003e\u003cspan address=\"10.3390/nu15112632\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGrotenhuis BA, Shapiro J, van Adrichem S, de Vries M, Koek M, Wijnhoven BP, van Lanschot JJB. Sarcopenia/Muscle Mass Is Not a Prognostic Factor for Short-and Long-Term Outcome After Esophagectomy for Cancer. World J Surg. 2016;40(11):2698\u0026ndash;704. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s00268-016-3603-1\u003c/span\u003e\u003cspan address=\"10.1007/s00268-016-3603-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePrado CM, Lieffers JR, McCargar LJ, Reiman T, Sawyer MB, Martin L, et al. Prevalence and clinical implications of sarcopenic obesity in patients with solid tumours of the respiratory and gastrointestinal tracts: a population-based study. Lancet Oncol. 2008;9(7):629\u0026ndash;35. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S1470-2045(08)70153-0\u003c/span\u003e\u003cspan address=\"10.1016/S1470-2045(08)70153-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMarra M, Sammarco R, De Lorenzo A, Iellamo F, Siervo M, et al. Assessment of Body Composition in Health and Disease Using Bioelectrical Impedance Analysis (BIA) and Dual Energy X-Ray Absorptiometry (DXA): A Critical Overview. Contrast Media Mol Imaging. 2019;2019:3548284. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1155/2019/3548284\u003c/span\u003e\u003cspan address=\"10.1155/2019/3548284\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLangius JAE, Kruizenga HM, van den Boogaardt J, et al. Comparison of the prognostic value of CT-determined sarcopenia and psoas muscle area in patients with advanced cancer. Support Care Cancer. 2018;26(5):1611\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMartin L, Birdsell L, Macdonald N, et al. Sarcopenia defined by measured appendicular lean mass is associated with postoperative infection and delayed recovery from colorectal cancer resection surgery. Br J Cancer. 2012;107(9):931\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKocak B, Baessler B, Bakas S, Cuocolo R, Fedorov A, Maier-Hein L, Mercaldo N, M\u0026uuml;ller H, Orlhac F, Pinto Dos Santos D, Stanzione A, Ugga L, Zwanenburg A. CheckList for EvaluAtion of Radiomics research (CLEAR): a step-by-step reporting guideline for authors and reviewers endorsed by ESR and EuSoMII. Insights Imaging. 2023;14(1):75. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13244-023-01415-8\u003c/span\u003e\u003cspan address=\"10.1186/s13244-023-01415-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKocak B, Akinci D'Antonoli T, Mercaldo N, et al. METhodological RadiomICs Score (METRICS): a quality scoring tool for radiomics research endorsed by EuSoMII. Insights into Imaging. 2024;15(1):8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13244-023-01572-w\u003c/span\u003e\u003cspan address=\"10.1186/s13244-023-01572-w\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 38228979; PMCID: PMC10792137.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePinto Almeida M, Fotsing G, Gijs E, Barigou M. Nutrition clinique: ce qui a chang\u0026eacute; en 2022 [Clinical nutrition: what's new in 2022]. Rev. Med. Suisse.2023;19(N\u0026deg;809\u0026thinsp;\u0026ndash;\u0026thinsp;10):46\u0026ndash;51. French. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.53738/REVMED.2023.19.809-10.46\u003c/span\u003e\u003cspan address=\"10.53738/REVMED.2023.19.809-10.46\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCruz-Jentoft AJ, Baeyens JP, Bauer JM, et al. Sarcopenia: European consensus on definition and diagnosis: report of the European Working Group on Sarcopenia in Older People. Age Ageing. 2010;39(4):412\u0026ndash;23. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/ageing/afq034\u003c/span\u003e\u003cspan address=\"10.1093/ageing/afq034\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMalmstrom TK, Miller DK, Simonsick EM, et al. SARC-F: a symptom score approach for the screening of sarcopenia. J Gerontol Biol Sci Med Sci. 2016;71(5):629\u0026ndash;34. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/gerona/glv215\u003c/span\u003e\u003cspan address=\"10.1093/gerona/glv215\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaracos VE, Martin L, Korc M, et al. Cancer-associated cachexia. Nat Rev Dis Primers. 2018;4:17105. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nrdp.2017.105\u003c/span\u003e\u003cspan address=\"10.1038/nrdp.2017.105\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMuscaritoli M, Anker SD, Argil\u0026eacute;s J, et al. Consensus definition of sarcopenia, cachexia and pre-cachexia: joint document elaborated by Special Interest Groups (SIG) cachexia-anorexia in chronic wasting diseases and nutrition in geriatrics. Clin Nutr. 2010;29(2):154\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.clnu.2009.12.004\u003c/span\u003e\u003cspan address=\"10.1016/j.clnu.2009.12.004\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAmini B, Boyle SP, Boutin RD, et al. Approaches to assessment of muscle mass and myosteatosis on computed tomography: a systematic review. Gerontol Biol Sci Med Sci. 2019;74(7):1079\u0026ndash;86. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/gerona/glz034\u003c/span\u003e\u003cspan address=\"10.1093/gerona/glz034\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJang MK, Park S, Raszewski R, et al. Prevalence and clinical implications of sarcopenia in breast cancer: a systematic review and meta-analysis. Support Care Cancer. 2024;32:328. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s00520-024-08532-0\u003c/span\u003e\u003cspan address=\"10.1007/s00520-024-08532-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKasahara K, Kono T, Sato Y, Ueno M, So H, Fuse Y, Shinden S, Ozawa H. Sarcopenia accompanied by systemic inflammation can predict clinical outcomes in patients with head and neck cancer undergoing curative therapy. Front Oncol. 2024;14:1378762. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fonc.2024.1378762\u003c/span\u003e\u003cspan address=\"10.3389/fonc.2024.1378762\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PMID: 38549928; PMCID: PMC10973154.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePrado CM, Heymsfield SB. Lean tissue imaging: a new era for nutritional assessment and intervention. JPEN J Parenter Enter Nutr. 2014;38(8):940\u0026ndash;53. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/0148607114550189\u003c/span\u003e\u003cspan address=\"10.1177/0148607114550189\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHe J, Huang Y, Huang N, Jiang J. Prevalence and predictive value of sarcopenia in surgically treated cholangiocarcinoma: a comprehensive review and meta-analysis. Front Oncol 14:1363843. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fonc.2024.1363843\u003c/span\u003e\u003cspan address=\"10.3389/fonc.2024.1363843\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGrotenhuis BA, Shapiro J, van Adrichem S, et al. Sarcopenia/muscle mass is not a prognostic factor for short-and long-term outcome after esophagectomy for cancer. World J Surg. 2016;40(7):2698\u0026ndash;704. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s00268-016-3603-1\u003c/span\u003e\u003cspan address=\"10.1007/s00268-016-3603-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen RH, Chang K, West RB. Clinical Applications of Machine Learning in Oncology. Nat Reviews Clin Oncol. 2018;15(9):568\u0026ndash;86. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41571-018-0004-z\u003c/span\u003e\u003cspan address=\"10.1038/s41571-018-0004-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang B, Tian J, Dong D, et al. Radiomics Features of Multiparametric MRI as Novel Prognostic Factors in Advanced Nasopharyngeal Carcinoma. Clin Cancer Res. 2017;23(15):4259\u0026ndash;69. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1158/1078-0432.CCR-16-2910\u003c/span\u003e\u003cspan address=\"10.1158/1078-0432.CCR-16-2910\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTables 1 to 11 are available in the Supplementary Files section.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmim","sideBox":"Learn more about [BMC Medical Imaging](http://bmcmedimaging.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmim/default.aspx","title":"BMC Medical Imaging","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"sarcopenia, CT radiomics, machine learning, prediction, gastric cancer","lastPublishedDoi":"10.21203/rs.3.rs-5350805/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5350805/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eObjectives\u003c/h2\u003e \u003cp\u003eSarcopenia, linked to postoperative survival in cancer patients, was investigated in this study. The research explored the relationship between CT imaging features of muscles in gastric cancer patients and their survival. Additionally, the study aimed to create a quantifiable survival prediction model using artificial intelligence.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eIn a retrospective study, 100 patients who underwent radical gastrectomy for gastric cancer were analyzed. After identifying sarcopenia using the psoas muscle index, clinical factors related to patient survival were investigated. Imaging features were extracted from manually delineated iliopsoas muscles and used in 11 machine learning algorithms. After completing the model training, we used a dataset comprising 34 patients from a secondary center as an external validation set to evaluate the model\u0026rsquo;s classification performance. After identifying the optimal model, we further explored the fusion methods of clinical omics and radiomics. Based on this, we constructed a predictive model for estimating the five-year survival rate of patients.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eClinical survival analysis highlighted age and tumor M stage as relevant factors. For the task of predicting five-year survival, we found that the Logistic Regression (LR) model without clinical feature fusion exhibited the most balanced and superior performance. Specifically, the AUC (Area Under Curve) values of this model on the training set, internal validation set, and external validation set were 0.82, 0.72, and 0.69, respectively. Additionally, the model\u0026rsquo;s accuracy remained relatively stable, approximately around 70%.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eIn this study, we developed a machine learning model based on preoperative CT imaging data of gastric cancer patients to predict their five-year survival rate. The model can achieve about 70% accuracy. Additionally, we explored the necessity and rationale of incorporating clinical independent factors into this predictive model. The results indicated a significant correlation between muscle imaging features and overall patient survival, highlighting the importance of sarcopenia in the clinical management of gastric cancer patients.\u003c/p\u003e","manuscriptTitle":"Predicting 5-Year Survival in Gastric Cancer Patients Using Iliopsoas Muscle CT Radiomics and Machine Learning Techniques","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-12-09 08:49:27","doi":"10.21203/rs.3.rs-5350805/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2024-11-11T16:14:10+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-11-06T15:18:18+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-11-06T15:16:58+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Imaging","date":"2024-10-29T03:40:23+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmim","sideBox":"Learn more about [BMC Medical Imaging](http://bmcmedimaging.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmim/default.aspx","title":"BMC Medical Imaging","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"52ed9f96-7280-4e6b-a96f-909fd59a79d0","owner":[],"postedDate":"December 9th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"in-revision","subjectAreas":[],"tags":[],"updatedAt":"2026-05-26T09:55:29+00:00","versionOfRecord":[],"versionCreatedAt":"2024-12-09 08:49:27","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5350805","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5350805","identity":"rs-5350805","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}