Journal of imaging informatics in medicine最新文献

{"title":"An Innovative Attention-based Triplet Deep Hashing Approach to Retrieve Histopathology Images.","authors":"Seyed Mohammad Alizadeh, Mohammad Sadegh Helfroush, M Emre Celebi","doi":"10.1007/s10278-024-01310-8","DOIUrl":"https://doi.org/10.1007/s10278-024-01310-8","url":null,"abstract":"<p><p>Content-based histopathology image retrieval (CBHIR) can assist in the diagnosis of different diseases. The retrieval procedure can be complex and time-consuming if high-dimensional features are required. Thus, hashing techniques are employed to address these issues by mapping the feature space into binary values of varying lengths. The performance of deep hashing approaches in image retrieval is often superior to that of traditional hashing methods. Among deep hashing approaches, triplet-based models are typically more effective than pairwise ones. Recent studies have demonstrated that incorporating the attention mechanism into a deep hashing approach can improve its effectiveness in retrieving images. This paper presents an innovative triplet deep hashing strategy based on the attention mechanism for retrieving histopathology images, called histopathology attention triplet deep hashing (HATDH). Three deep attention-based hashing models with identical architectures and weights are employed to produce binary values. The proposed attention module can aid the models in extracting features more efficiently. Moreover, we introduce an improved triplet loss function considering pair inputs separately in addition to triplet inputs for increasing efficiency during the training and retrieval steps. Based on experiments conducted on two public histopathology datasets, BreakHis and Kather, HATDH significantly outperforms state-of-the-art hashing algorithms.</p>","PeriodicalId":516858,"journal":{"name":"Journal of imaging informatics in medicine","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142635734","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Preoperative Prediction of Perineural Invasion in Pancreatic Ductal Adenocarcinoma Using Machine Learning Radiomics Based on Contrast-Enhanced CT Imaging.","authors":"Wenzheng Lu, Yanqi Zhong, Xifeng Yang, Yuxi Ge, Heng Zhang, Xingbiao Chen, Shudong Hu","doi":"10.1007/s10278-024-01325-1","DOIUrl":"https://doi.org/10.1007/s10278-024-01325-1","url":null,"abstract":"<p><p>The objective of the study is to assess the clinical value of machine learning radiomics based on contrast-enhanced computed tomography (CECT) images in preoperative prediction of perineural invasion (PNI) status in pancreatic ductal adenocarcinoma (PDAC). A total of 143 patients with PDAC were enrolled in this retrospective study (training group, n = 100; test group, n = 43). Radiomics features were extracted from CECT images and selected by the Mann-Whitney U-test, Pearson correlation coefficient, and least absolute shrinkage and selection operator (LASSO). The logistic regression (LR), support vector machine (SVM), random forest (RF), extreme gradient boosting (XGBoost), and decision tree (DT) algorithms were trained to build radiomics models by radiomic features. Multivariate logistic regression was employed to identify independent predictors and establish clinical models. A combined model was constructed by integrating clinical and radiomics features. Model performances were assessed by receiver operating characteristic curves (ROCs) and decision curve analyses (DCAs). A total of 788 radiomics features were extracted from CECT images, of which 14 were identified as significant through the three-step selection process. Among the machine learning models, the SVM radiomics model exhibited the highest predictive performance in the test group, achieving an area under the curve (AUC) of 0.831, accuracy of 0.698, sensitivity of 0.677, and specificity of 0.750. After logistic regression screening, the clinical model was established using carbohydrate antigen 19-9 (CA199) levels as one independent predictor. In the test group, the clinical model demonstrated an AUC of 0.644, accuracy of 0.744, sensitivity of 0.871, and specificity of 0.417. The combined model showed improved performance compared to both the clinical and radiomics models in the test group, with an AUC of 0.844, accuracy of 0.767, sensitivity of 0.806, and specificity of 0.667. Subsequently, DCA of the combined model indicated optimal clinical value for predicting PNI status. Machine learning radiomics models can accurately predict PNI status in patients with pancreatic ductal adenocarcinoma. The combined model, which incorporates clinical and radiomics features, enhances preoperative diagnostic performance and aids in the selection of treatment methods.</p>","PeriodicalId":516858,"journal":{"name":"Journal of imaging informatics in medicine","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142633166","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Robust Radiomics Models for Predicting HIFU Prognosis in Uterine Fibroids Using SHAP Explanations: A Multicenter Cohort Study.","authors":"Huan Liu, Jincheng Zeng, Chen Jinyun, Xiaohua Liu, Yongbin Deng, Chenghai Li, Faqi Li","doi":"10.1007/s10278-024-01318-0","DOIUrl":"https://doi.org/10.1007/s10278-024-01318-0","url":null,"abstract":"<p><p>This study sought to develop and validate different machine learning (ML) models that leverage non-contrast MRI radiomics to predict the degree of nonperfusion volume ratio (NVPR) of high-intensity focused ultrasound (HIFU) treatment for uterine fibroids, equipping clinicians with an early prediction tool for decision-making. This study conducted a retrospective analysis on 221 patients with uterine fibroids who received HIFU treatment and were divided into a training set (N = 117), internal validation (N = 49), and an external test set (N = 55). The 851 radiomics features were extracted from T2-weighted imaging (T2WI), and the max-relevance and min-redundancy (mRMR) and the least absolute shrinkage and selection operator (LASSO) regression were applied for feature selection. Several ML models were constructed by logistic regression (LR), decision tree (DT), random forest (RF), support vector machine (SVM), extreme gradient boosting (XGBoost), and light gradient boosting machine (LGBM). These models underwent internal and external validation, and the best model's feature significance was assessed via the Shapley additive explanations (SHAP) method. Four significant non-contrast MRI radiomics features were identified, with the SVM model outperforming others in both internal and external validations, and the AUCs of the T2WI models were 0.860, 0.847, and 0.777, respectively. SHAP analysis highlighted five critical predictors of postoperative NVPR degree, encompassing two radiomics features from non-contrast MRI and three clinical data indicators. The SVM model combining radiomics features and clinical parameters effectively predicts NVPR degree post-HIFU, which enables timely and effective interventions of HIFU.</p>","PeriodicalId":516858,"journal":{"name":"Journal of imaging informatics in medicine","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142634782","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An Automated Diagnosis of Parkinson's Disease from MRI Scans Based on Enhanced Residual Dense Network with Attention Mechanism.","authors":"Hakan Acikgoz, Deniz Korkmaz, Tarık Talan","doi":"10.1007/s10278-024-01316-2","DOIUrl":"https://doi.org/10.1007/s10278-024-01316-2","url":null,"abstract":"<p><p>The increasing prevalence of neurodegenerative diseases has recently heightened interest in research on early diagnosis of these diseases. Parkinson's disease (PD), among the most prominent of these conditions, is a neurological disorder causing the loss of nerve cells and significantly affecting movement control. Detection of PD in early stages is of critical importance to prevent the progression of the disease and improve treatment processes. The aim of the current study is to develop a deep learning model that can perform accurate classification for early diagnosis of PD from MRI images. In this study, a densely connected feature fusion network with residual learning is designed to diagnose PD patients. The designed network consists of a serial dense block with skip connections and efficient attention mechanisms. In this architecture, squeeze-excitation (SE) blocks with ResNeXt (SE-ResNeXt block) modules are utilized to extract distinctive and high-level features. In the experiments, a publicly available T2-weighted MRI dataset is used, and an offline augmentation process is applied to limited data to increase the generalization ability and classification performance. The proposed method is evaluated and compared with current state-of-the-art deep learning methods. The obtained results show that the proposed model gives higher classification performance with an overall accuracy of 94.44%, precision of 91.67%, sensitivity of 91.67%, specificity of 95.83%, F1-score of 91.67%, and Matthew's correlation coefficient of 87.50% for the PD and healthy control subjects.</p>","PeriodicalId":516858,"journal":{"name":"Journal of imaging informatics in medicine","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142635733","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Comparative Evaluation of Large Language Model Utility in Neuroimaging Clinical Decision Support.","authors":"Luke Miller, Peter Kamel, Jigar Patel, Jay Agrawal, Min Zhan, Nathan Bumbarger, Kenneth Wang","doi":"10.1007/s10278-024-01161-3","DOIUrl":"https://doi.org/10.1007/s10278-024-01161-3","url":null,"abstract":"<p><p>Imaging utilization has increased dramatically in recent years, and at least some of these studies are not appropriate for the clinical scenario. The development of large language models (LLMs) may address this issue by providing a more accessible reference resource for ordering providers, but their relative performance is currently understudied. Evaluate and compare the relative appropriateness and usefulness of imaging recommendations generated by eight publicly available models in response to neuroradiology clinical scenarios. Twenty-four common neuroradiology clinical scenarios were selected which often yield suboptimal imaging utilization. Questions were crafted to assess the ability of LLMs to provide accurate and actionable advice. The LLMs were assessed in August 2023 using natural-language 1-2 sentence queries requesting advice about optimal image ordering given certain clinical parameters. Eight of the most well-known LLMs were chosen for evaluation: ChatGPT, GPT4, Bard (Versions 1 and 2), Bing Chat, Llama 2, Perplexity, and Claude. The models were graded by three fellowship-trained neuroradiologists on whether their advice was \"optimal\" or \"not optimal\" according to the ACR Appropriateness Criteria or the New Orleans Head CT Criteria. The raters also ranked the models based on the appropriateness, helpfulness, concision, and source-citations in their response. The models varied in their ability to deliver an \"optimal\" recommendation based on these scenarios as follows: ChatGPT (20/24), GPT4 (23/24), Bard 1 (13/24), Bard 2 (14/24), Bing Chat (14/24), Llama (5/24), Perplexity (19/24), and Claude (19/24). The median ranks of the LLMs were as follows: ChatGPT (3), GPT4 (1.5), Bard 1 (4.5), Bard 2 (5), Bing Chat (6), Llama (7.5), Perplexity (4), and Claude (3). Characteristic errors are described and discussed. GPT-4, ChatGPT, and Claude generally outperformed Bard, Bing Chat, and Llama 2. This study evaluates the performance of a greater variety of publicly available LLMs in settings that more closely mimic real-world use cases as well as discussing the practical challenges of doing so. This is the first study to evaluate and compare a wide range of publicly available LLMs to determine appropriateness of their neuroradiology imaging recommendations.</p>","PeriodicalId":516858,"journal":{"name":"Journal of imaging informatics in medicine","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142604550","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Cone-Beam CT to CT Image Translation Using a Transformer-Based Deep Learning Model for Prostate Cancer Adaptive Radiotherapy.","authors":"Yuhei Koike, Hideki Takegawa, Yusuke Anetai, Satoaki Nakamura, Ken Yoshida, Asami Yoshida, Midori Yui, Kazuki Hirota, Kenichi Ueda, Noboru Tanigawa","doi":"10.1007/s10278-024-01312-6","DOIUrl":"https://doi.org/10.1007/s10278-024-01312-6","url":null,"abstract":"<p><p>Cone-beam computed tomography (CBCT) is widely utilized in image-guided radiation therapy; however, its image quality is poor compared to planning CT (pCT), thus restricting its utility for adaptive radiotherapy (ART). Our objective was to enhance CBCT image quality utilizing a transformer-based deep learning model, SwinUNETR, which we compared with a conventional convolutional neural network (CNN) model, U-net. This retrospective study involved 260 patients undergoing prostate radiotherapy, with 245 patients used for training and 15 patients reserved as an independent hold-out test dataset. Employing a CycleGAN framework, we generated synthetic CT (sCT) images from CBCT images, employing SwinUNETR and U-net as generators. We evaluated sCT image quality and assessed its dosimetric impact for photon therapy through gamma analysis and dose-volume histogram (DVH) comparisons. The mean absolute error values for the CT numbers, calculated using all voxels within the patient's body contour and taking the pCT images as a reference, were 84.07, 73.49, and 64.69 Hounsfield units for CBCT, U-net, and SwinUNETR images, respectively. Gamma analysis revealed superior agreement between the dose on the pCT images and on the SwinUNETR-based sCT plans compared to those based on U-net. DVH parameters calculated on the SwinUNETR-based sCT deviated by < 1% from those in pCT plans. Our study showed that, compared to the U-net model, SwinUNETR could proficiently generate more precise sCT images from CBCT images, facilitating more accurate dose calculations. This study demonstrates the superiority of transformer-based models over conventional CNN-based approaches for CBCT-to-CT translation, contributing to the advancement of image synthesis techniques in ART.</p>","PeriodicalId":516858,"journal":{"name":"Journal of imaging informatics in medicine","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142606222","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Diagnostic Performance of a Next-Generation Virtual/Augmented Reality Headset: A Pilot Study of Diverticulitis on CT.","authors":"Paul M Murphy, Julie Y An, Luke M Wojdyla, Adam C Searleman, Aman Khurana, Thomas W Loehfelm, Kathryn J Fowler, Christopher A Longhurst","doi":"10.1007/s10278-024-01292-7","DOIUrl":"https://doi.org/10.1007/s10278-024-01292-7","url":null,"abstract":"<p><p>Next-generation virtual/augmented reality (VR/AR) headsets may rival the desktop computer systems that are approved for clinical interpretation of radiologic images, but require validation for high-resolution low-luminance diagnoses like diverticulitis. The primary aim of this study is to compare diagnostic performance for detecting diverticulitis on CT between radiologists using a headset versus a desktop. The secondary aim is to survey participating radiologists about the usage of both devices. This pilot study retrospectively included 110 patients (mean age 64 ± 14 years, 62 women) who had abdomen/pelvis CT scans for which the report mentioned the presence or absence of diverticulitis. Scans were dichotomized and matched by time, for a total of 55 cases with diverticulitis and 55 controls with no diverticulitis. Six radiologists were oriented to the VR/AR headset (Apple Vision Pro) and viewer app (Visage Ease VP) using ten scans. They each scored 100 unknown scans on a 6-level scale for diverticulitis (1 = no diverticulitis, 6 = diverticulitis) on the headset and then on a desktop. Time per case was recorded. Finally, they completed a survey using 5-level scales about the ease of use of the headset and viewer app (1 = difficult, 5 = easy), about their experience with the headset (1 = bad, 5 = good), and about their preference between devices (1 = desktop, 5 = headset). Summary statistics and multi-reader multi-case ROC curves were calculated. The AUC (and 95% confidence interval) for diverticulitis was 0.93 (0.88-0.97) with the headset and 0.94 (0.91-0.98) with the desktop (p = 0.40). The median (and first-third quartiles) of time per case was 57 (41-76) seconds for the headset and 31 (22-64) seconds for the desktop (p < 0.001). Average survey scores ranged from 3.3 to 5 for ease of use, from 3 to 4.7 for experience, and from 2.2 to 3.3 for preference. Diagnostic performance for detecting diverticulitis on CT was similar between the next-generation VR/AR headset and desktop. Ease of use, experience, and preference varied across different aspects of the devices and among radiologists.</p>","PeriodicalId":516858,"journal":{"name":"Journal of imaging informatics in medicine","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142570883","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Volumetric Integrated Classification Index: An Integrated Voxel-Based Morphometry and Machine Learning Interpretable Biomarker for Post-Traumatic Stress Disorder.","authors":"Yulong Jia, Beining Yang, Haotian Xin, Qunya Qi, Yu Wang, Liyuan Lin, Yingying Xie, Chaoyang Huang, Jie Lu, Wen Qin, Nan Chen","doi":"10.1007/s10278-024-01313-5","DOIUrl":"https://doi.org/10.1007/s10278-024-01313-5","url":null,"abstract":"<p><p>PTSD is a complex mental health condition triggered by individuals' traumatic experiences, with long-term and broad impacts on sufferers' psychological health and quality of life. Despite decades of research providing partial understanding of the pathobiological aspects of PTSD, precise neurobiological markers and imaging indicators remain challenging to pinpoint. This study employed VBM analysis and machine learning algorithms to investigate structural brain changes in PTSD patients. Data were sourced ADNI-DoD database for PTSD cases and from the ADNI database for healthy controls. Various machine learning models, including SVM, RF, and LR, were utilized for classification. Additionally, the VICI was proposed to enhance model interpretability, incorporating SHAP analysis. The association between PTSD risk genes and VICI values was also explored through gene expression data analysis. Among the tested machine learning algorithms, RF emerged as the top performer, achieving high accuracy in classifying PTSD patients. Structural brain abnormalities in PTSD patients were predominantly observed in prefrontal areas compared to healthy controls. The proposed VICI demonstrated classification efficacy comparable to the optimized RF model, indicating its potential as a simplified diagnostic tool. Analysis of gene expression data revealed significant associations between PTSD risk genes and VICI values, implicating synaptic integrity and neural development regulation. This study reveals neuroimaging and genetic characteristics of PTSD, highlighting the potential of VBM analysis and machine learning models in diagnosis and prognosis. The VICI offers a promising approach to enhance model interpretability and guide clinical decision-making. These findings contribute to a better understanding of the pathophysiological mechanisms of PTSD and provide new avenues for future diagnosis and treatment.</p>","PeriodicalId":516858,"journal":{"name":"Journal of imaging informatics in medicine","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142577395","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Learnable Context in Multiple Instance Learning for Whole Slide Image Classification and Segmentation.","authors":"Yu-Yuan Huang, Wei-Ta Chu","doi":"10.1007/s10278-024-01302-8","DOIUrl":"https://doi.org/10.1007/s10278-024-01302-8","url":null,"abstract":"<p><p>Multiple instance learning (MIL) has become a cornerstone in whole slide image (WSI) analysis. In this paradigm, a WSI is conceptualized as a bag of instances. Instance features are extracted by a feature extractor, and then a feature aggregator fuses these instance features into a bag representation. In this paper, we advocate that both feature extraction and aggregation can be enhanced by considering the context or correlation between instances. We learn contextual features between instances, and then fuse contextual features with instance features to enhance instance representations. For feature aggregation, we observe performance instability particularly when disease-positive instances are only a minor fraction of the WSI. We introduce a self-attention mechanism to discover correlation among instances and foster more effective bag representations. Through comprehensive testing, we have demonstrated that the proposed method outperforms existing WSI classification methods by 1 to 4% classification accuracy, based on the Camelyon16 and the TCGA-NSCLC datasets. The proposed method also outperforms the most recent weakly supervised WSI segmentation method by 0.6 in terms of the Dice coefficient, based on the Camelyon16 dataset.</p>","PeriodicalId":516858,"journal":{"name":"Journal of imaging informatics in medicine","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142570886","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Applying Deep-Learning Algorithm Interpreting Kidney, Ureter, and Bladder (KUB) X-Rays to Detect Colon Cancer.","authors":"Ling Lee, Chin Lin, Chia-Jung Hsu, Heng-Hsiu Lin, Tzu-Chiao Lin, Yu-Hong Liu, Je-Ming Hu","doi":"10.1007/s10278-024-01309-1","DOIUrl":"https://doi.org/10.1007/s10278-024-01309-1","url":null,"abstract":"<p><p>Early screening is crucial in reducing the mortality of colorectal cancer (CRC). Current screening methods, including fecal occult blood tests (FOBT) and colonoscopy, are primarily limited by low patient compliance and the invasive nature of the procedures. Several advanced imaging techniques such as computed tomography (CT) and histological imaging have been integrated with artificial intelligence (AI) to enhance the detection of CRC. There are still limitations because of the challenges associated with image acquisition and the cost. Kidney, ureter, and bladder (KUB) radiograph which is inexpensive and widely used for abdominal assessments in emergency settings and shows potential for detecting CRC when enhanced using advanced techniques. This study aimed to develop a deep learning model (DLM) to detect CRC using KUB radiographs. This retrospective study was conducted using data from the Tri-Service General Hospital (TSGH) between January 2011 and December 2020, including patients with at least one KUB radiograph. Patients were divided into development (n = 28,055), tuning (n = 11,234), and internal validation (n = 16,875) sets. An additional 15,876 patients were collected from a community hospital as the external validation set. A 121-layer DenseNet convolutional network was trained to classify KUB images for CRC detection. The model performance was evaluated using receiver operating characteristic curves, with sensitivity, specificity, and area under the curve (AUC) as metrics. The AUC, sensitivity, and specificity of the DLM in the internal and external validation sets achieved 0.738, 61.3%, and 74.4%, as well as 0.656, 47.7%, and 72.9%, respectively. The model performed better for high-grade CRC, with AUCs of 0.744 and 0.674 in the internal and external sets, respectively. Stratified analysis showed superior performance in females aged 55-64 with high-grade cancers. AI-positive predictions were associated with a higher long-term risk of all-cause mortality in both validation cohorts. AI-enhanced KUB X-ray analysis can enhance CRC screening coverage and effectiveness, providing a cost-effective alternative to traditional methods. Further prospective studies are necessary to validate these findings and fully integrate this technology into clinical practice.</p>","PeriodicalId":516858,"journal":{"name":"Journal of imaging informatics in medicine","volume":" ","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-10-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142560009","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}