European Journal of Radiology最新文献

{"title":"Corrigendum to “Prospective study of dual-phase 99mTc-MIBI SPECT/CT nomogram for differentiating non-small cell lung cancer from benign pulmonary lesions” [Eur. J. Radiol. 179 (2024) 111657]","authors":"","doi":"10.1016/j.ejrad.2024.111704","DOIUrl":"10.1016/j.ejrad.2024.111704","url":null,"abstract":"","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S0720048X24004200/pdfft?md5=aefc464d7daad20673240c6ead80e2b2&pid=1-s2.0-S0720048X24004200-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142145470","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Quantification of bone microarchitecture using photon-counting CT at different radiation doses: A comparison with µCT","authors":"","doi":"10.1016/j.ejrad.2024.111717","DOIUrl":"10.1016/j.ejrad.2024.111717","url":null,"abstract":"<div><h3>Purpose</h3><p>Accurate measurements of trabecular bone microarchitecture are required for a proper assessment of bone fragility. Photon-counting detector CT (PCD-CT) has different technical properties than conventional CT, resulting in higher resolution and thereby potentially enabling <em>in-vivo</em> measurement of trabecular microarchitecture. The purpose of this study was to quantify trabecular bone microarchitectural parameters with PCD-CT at varying radiation doses and compare this to µCT as gold standard.</p></div><div><h3>Method</h3><p>Both distal radii, distal tibiae, femoral heads, and two vertebrae were dissected from one human. All specimens were scanned <em>ex-vivo</em> on a PCD-CT system (slice increment 0.1 mm; pixel size 0.1042–0.127 mm) and a µCT system (isotropic voxel size 49–68.4 µm). The radiation doses of the PCD-CT scans were varied from 2.5 to 120 mGy based on the volume CT dose index (CTDI_vol32). For the PCD-CT scans, contrast-to-noise ratio and trabecular sharpness were calculated and compared between radiation doses. µCT and PCD-CT scans were registered. The trabecular bone was then segmented from all PCD-CT and µCT scans and split into cubes with 6-mm edge length. For each cube, bone volume over total volume, trabecular thickness, trabecular number, and trabecular heterogeneity were calculated and compared between corresponding PCD-CT and µCT cubes.</p></div><div><h3>Results</h3><p>With increasing dose, contrast-to-noise ratio and trabecular sharpness values increased for the PCD-CT images. Already at the lowest dose, high correlations between the trabecular microarchitectural parameters between µCT and PCD-CT were found (R² = 0.55–0.95), which improved with increasing radiation dose (R² = 0.76–0.96 at 20 mGy).</p></div><div><h3>Conclusions</h3><p>PCD-CT can be used to quantify trabecular bone microarchitecture, with accuracy comparable to µCT and at clinically relevant radiation doses.</p></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S0720048X24004339/pdfft?md5=707c0b792b17a6ee84bf01abcc85f796&pid=1-s2.0-S0720048X24004339-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142145455","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Prognostic implication of extra-pancreatic organ invasion in resectable pancreas ductal adenocarcinoma in the pancreas tail","authors":"","doi":"10.1016/j.ejrad.2024.111715","DOIUrl":"10.1016/j.ejrad.2024.111715","url":null,"abstract":"<div><h3>Objectives</h3><p>To assess the prognostic significance of extra-pancreatic organ invasion in patients with resectable pancreatic ductal adenocarcinoma (PDAC) in the pancreas tail.</p></div><div><h3>Materials &amp; Methods</h3><p>This retrospective study included patients with resectable PDAC in the pancreas tail who received upfront surgery between 2014 and 2020 at a tertiary institution. Preoperative pancreas protocol computed tomography (CT) scans evaluated tumor size, peripancreatic tumor infiltration, suspicious metastatic lymph nodes, and extra-pancreatic organ invasion. The influence of extra-pancreatic organ invasion, detected by CT or postoperative pathology, on pathologic resection margin status was evaluated using logistic regression. The impact on recurrence-free survival (RFS) was analyzed using multivariable Cox proportional hazard models (clinical-CT and clinical-pathologic).</p></div><div><h3>Results</h3><p>The study included 158 patients (mean age, 65 years ± 8.8 standard deviation; 93 men). Extra-pancreatic organ invasion identified by either CT (<em>p</em> = 0.92) or pathology (<em>p</em> = 0.99) was not associated with a positive resection margin. Neither CT (<em>p</em> = 0.42) nor pathological (<em>p</em> = 0.64) extra-pancreatic organ invasion independently correlated with RFS. Independent predictors for RFS included suspicious metastatic lymph node (hazard ratio [HR], 2.05; 95 % confidence interval [CI], 1.08–3.9; <em>p</em> = 0.03) on CT in the clinical-CT model, pathological T stage (HR, 2.97; 95 % confidence interval [CI], 1.39–6.35; <em>p</em> = 0.005 for T2 and HR, 3.78; 95 % CI, 1.64–8.76; <em>p</em> = 0.002 for T3) and adjuvant therapy (HR, 0.62; 95 % confidence interval [CI], 0.42–0.92; <em>p</em> = 0.02) in the clinical-pathologic model.</p></div><div><h3>Conclusion</h3><p>Extra-pancreatic organ invasion does not independently influence pathologic resection margin status and RFS in patients with resectable PDAC in the pancreas tail after curative-intent resection; therefore, it should not be considered a high-risk factor.</p></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142145475","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessment of the contribution of the ADC value to the Kaiser score in the differential diagnosis of breast lesions with non-mass enhancement morphology on MRI","authors":"","doi":"10.1016/j.ejrad.2024.111713","DOIUrl":"10.1016/j.ejrad.2024.111713","url":null,"abstract":"<div><h3>Purpose</h3><p>To investigate the effectiveness of diffusion-weighted imaging (DWI) as a supplementary tool to the Kaiser score (KS) in diagnosing breast cancer in non-mass enhancement (NME) lesions using breast magnetic resonance imaging (MRI).</p></div><div><h3>Methods</h3><p>This single-center, retrospective study analyzed 360 cases with NME on MRI images. Two breast radiologists independently evaluated each lesion using the Kaiser score (KS) and apparent diffusion coefficient (ADC) values, without knowledge of the pathological outcomes. NME lesions with a KS above 4 and an ADC value below 1.3 × 10^-3mm²/s were classified as malignant. Inter-rater reliability was determined using Cohen’s Kappa (κ) statistics. The diagnostic performance of KS, DWI, and their combination was assessed by calculating sensitivity, specificity, and the area under the curve (AUC), and the results were compared across the benign and malignant groups.</p></div><div><h3>Results</h3><p>The diagnostic performance of KS surpassed that of DWI in predicting the malignancy of NMEs (p = 0.003). The sensitivity of KS alone was 93 %; however, when ADC data was incorporated, the sensitivity decreased to 86 %, with no significant difference observed (p = 0.060). The specificity of the combined KS and ADC (94 %) was significantly higher than that of KS alone (89 %) and DWI alone (73 %) (p &lt; 0.001).</p></div><div><h3>Conclusion</h3><p>Our findings indicated that although the combination of KS and ADC increased specificity and reduced unnecessary biopsies, the resulting decrease in sensitivity was unacceptable. Therefore, KS alone is superior to the KS-ADC combination in detecting malignancy in NME lesions.</p></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142145469","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine learning value in the diagnosis of vertebral fractures: A systematic review and meta-analysis","authors":"","doi":"10.1016/j.ejrad.2024.111714","DOIUrl":"10.1016/j.ejrad.2024.111714","url":null,"abstract":"<div><h3>Purpose</h3><p>To evaluate the diagnostic accuracy of machine learning (ML) in detecting vertebral fractures, considering varying fracture classifications, patient populations, and imaging approaches.</p></div><div><h3>Method</h3><p>A systematic review and meta-analysis were conducted by searching PubMed, Embase, Cochrane Library, and Web of Science up to December 31, 2023, for studies using ML for vertebral fracture diagnosis. Bias risk was assessed using QUADAS-2. A bivariate mixed-effects model was used for the meta-analysis. Meta-analyses were performed according to five task types (vertebral fractures, osteoporotic vertebral fractures, differentiation of benign and malignant vertebral fractures, differentiation of acute and chronic vertebral fractures, and prediction of vertebral fractures). Subgroup analyses were conducted by different ML models (including ML and DL) and modeling methods (including CT, X-ray, MRI, and clinical features).</p></div><div><h3>Results</h3><p>Eighty-one studies were included. ML demonstrated a diagnostic sensitivity of 0.91 and specificity of 0.95 for vertebral fractures. Subgroup analysis showed that DL (SROC 0.98) and CT (SROC 0.98) performed best overall. For osteoporotic fractures, ML showed a sensitivity of 0.93 and specificity of 0.96, with DL (SROC 0.99) and X-ray (SROC 0.99) performing better. For differentiating benign from malignant fractures, ML achieved a sensitivity of 0.92 and specificity of 0.93, with DL (SROC 0.96) and MRI (SROC 0.97) performing best. For differentiating acute from chronic vertebral fractures, ML showed a sensitivity of 0.92 and specificity of 0.93, with ML (SROC 0.96) and CT (SROC 0.97) performing best. For predicting vertebral fractures, ML had a sensitivity of 0.76 and specificity of 0.87, with ML (SROC 0.80) and clinical features (SROC 0.86) performing better.</p></div><div><h3>Conclusions</h3><p>ML, especially DL models applied to CT, MRI, and X-ray, shows high diagnostic accuracy for vertebral fractures. ML also effectively predicts osteoporotic vertebral fractures, aiding in tailored prevention strategies. Further research and validation are required to confirm ML’s clinical efficacy.</p></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S0720048X24004303/pdfft?md5=f764b2ad4d25f2e6a6cc2fa1c1f5907e&pid=1-s2.0-S0720048X24004303-main.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142145473","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Comparative analysis of hepatic fat quantification across 5 T, 3 T and 1.5 T: A study on consistency and feasibility","authors":"","doi":"10.1016/j.ejrad.2024.111709","DOIUrl":"10.1016/j.ejrad.2024.111709","url":null,"abstract":"<div><h3>Objectives</h3><p>Magnetic resonance imaging (MRI) is a critical noninvasive technique for evaluating liver steatosis, with efficient and precise fat quantification being essential for diagnosing liver diseases. This study leverages 5 T ultra-high-field MRI to demonstrate the clinical significance of liver fat quantification, and explores the consistency and accuracy of the Proton Density Fat Fraction (PDFF) in the liver across different magnetic field strengths and measurement methodologies.</p></div><div><h3>Methods</h3><p>The study involved phantoms with lipid contents ranging from 0 % to 30 % and 35 participants (21 females, 14 males; average age 30.17 ± 13.98 years, body mass index 25.84 ± 4.76, waist-hip ratio 0.84 ± 0.09). PDFF measurements were conducted using chemical shift encoded (CSE) MRI at 5 T, 3 T, and 1.5 T, alongside magnetic resonance spectroscopy (MRS) at 5 T and 1.5 T for both liver and phantoms, analyzed using jMRUI software. The MRS-derived PDFF values served as the reference standard. Repeatability of 5 T MRI measurements was assessed through correlation analysis, while accuracy was evaluated using linear regression analysis against the reference standards.</p></div><div><h3>Results</h3><p>The CSE-PDFF measurements at 5 T demonstrated strong consistency with those at 3 T and 1.5 T, showing high intraclass correlation coefficients (ICC) of 0.988 and 0.980, respectively (all p &lt; 0.001). There was also significant consistency across ROIs within liver lobes, with ICC values ranging from 0.975 to 0.986 (all p &lt; 0.001). MRS-PDFF measurements for both phantoms and liver at 5 T and 1.5 T exhibited substantial agreement, with ICC values of 0.996 and 0.980, respectively (all p &lt; 0.001). Particularly, ICC values for ROIs in the liver ranged from 0.963 to 0.990 (all p &lt; 0.001). Despite overall agreement, statistically significant differences were noted in specific ROIs within the liver lobes (p = 0.004 and 0.012). The CSE and MRS PDFF measurements at 5 T displayed strong consistency, with an ICC of 0.988 (p &lt; 0.001), and significant agreement was also found between 5 T CSE and 1.5 T MRS PDFF measurements, with an ICC of 0.978 (p &lt; 0.001). Agreement was significant within the ROIs of the liver lobes on the same platform at 5 T, with ICC values ranging from 0.986 to 0.991 (all p &lt; 0.001).</p></div><div><h3>Conclusion</h3><p>PDFF measurements at 5 T MR imaging exhibited both accuracy and repeatability, indicating that 5 T imaging provides reliable quantification of liver fat content and shows substantial potential for clinical diagnostic applications.</p></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-08-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142117729","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"PSMA PET improves characterization of dural-based intracranial lesions in patients with metastatic prostate cancer","authors":"","doi":"10.1016/j.ejrad.2024.111711","DOIUrl":"10.1016/j.ejrad.2024.111711","url":null,"abstract":"<div><h3>Purpose</h3><p>Theranostic approaches combining prostate-specific membrane antigen (PSMA)-PET/CT or PET/MRI with PSMA-targeted radionuclide therapy have improved clinical outcomes in patients with prostate cancer (PCa) especially metastatic castrate resistant prostate cancer. Dural metastases in PCa are rare but can pose a diagnostic challenge, as meningiomas, a more common dural based lesions have been shown to express PSMA. The aim of this study is to compare PSMA PET parameters between brain lesions classified as dural metastases and meningiomas in prostate cancer patients.</p></div><div><h3>Methods</h3><p>A retrospective analysis of PSMA PET/CT scans in patients with PCa and intracranial lesions was conducted. Brain lesions were categorized as dural metastases or meningiomas based on MRI characteristics, longitudinal follow-up, and histopathological characteristics. Standardized uptake values (SUVmax) of each brain lesion were measured, along with SUV ratio referencing parotid gland (SUVR). SUVs between lesions classified as metastases and meningiomas, respectively, were compared using Mann-Whitney-test. Diagnostic accuracy was evaluated using ROC analysis.</p></div><div><h3>Results</h3><p>26 male patients (median age: 76.5 years, range: 59–96 years) met inclusion criteria. A total of 44 lesions (7 meningiomas and 37 metastases) were analyzed. Median SUVmax and SUVR were significantly lower in meningiomas compared to metastases (SUVmax: 2.7 vs. 11.5, p = 0.001; SUVR: 0.26 vs. 1.05, p &lt; 0.001). ROC analysis demonstrated AUC 0.903; the optimal cut-off value for SUVR was 0.81 with 81.1 % sensitivity and 100 % specificity.</p></div><div><h3>Conclusion</h3><p>PSMA PET has the potential to differentiate meningiomas from dural-based metastases in patients with PCa, which can optimize clinical management and thus improve patient outcomes.</p></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-08-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142121842","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Head-to-head comparison of contrast-enhanced CT, dual-layer spectral-detector CT, and Gd-EOB-DTPA-enhanced MR in detecting neuroendocrine tumor liver metastases","authors":"","doi":"10.1016/j.ejrad.2024.111710","DOIUrl":"10.1016/j.ejrad.2024.111710","url":null,"abstract":"<div><h3>Purpose</h3><p>To explore the optimal of kiloelectron voltage (keV) of virtual monoenergetic imaging (VMI) of dual-layer spectral-detector CT (DLCT) in detecting neuroendocrine tumor liver metastases (NETLM) and to investigate diagnostic performance of polyenergetic images (PEI), DLCT, and Gd-EOB-DTPA-enhanced MR.</p></div><div><h3>Methods</h3><p>Seventy-two patients with suspected NETLM who underwent DLCT and Gd-EOB-DTPA-enhanced MR were retrospectively enrolled. Tumor signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were compared between PEI and VMI at 40–140 keV. Two radiologists read the CT examinations with and without VMI separately in consensus. Two other radiologists read the Gd-EOB-DTPA-enhanced MR in consensus. The diagnostic performance was evaluated. Reference standard was histopathology, follow-up, and interpretation of all available imaging.</p></div><div><h3>Results</h3><p>The highest SNR and CNR were observed at VMI40_keV, significantly higher than PEI in the arterial and venous phases (all <em>P</em>&lt;0.01). A total of 477 lesions were identified (396 metastases, 81 benign lesions). Per-lesion AUC was 0.86, 0.91, and 0.97 (PEI, DLCT, and Gd-EOB-DTPA-enhanced MR, respectively). Sensitivity of PEI, DLCT, and Gd-EOB-DTPA-enhanced MRI were 0.76, 0.86, and 0.95, respectively. DLCT significantly improved sensitivity compared to PEI. MR had significantly higher sensitivity than DLCT and PEI. Subgroup analysis demonstrated that the difference in diagnostic performance was concentrated on lesions &lt; 10 mm.</p></div><div><h3>Conclusion</h3><p>The image quality of VMI40_keV is higher than that of PEI. DLCT with VMI40_keV provides better diagnostic sensitivity for NETLM detection than PEI. Gd-EOB-DTPA-enhanced MR yielded the best diagnostic performance for NETLM detection.</p></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-08-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142145472","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Beyond accuracy: Reproducibility must lead AI advances in radiology","authors":"","doi":"10.1016/j.ejrad.2024.111703","DOIUrl":"10.1016/j.ejrad.2024.111703","url":null,"abstract":"","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142095415","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"2.5D deep learning based on multi-parameter MRI to differentiate primary lung cancer pathological subtypes in patients with brain metastases","authors":"","doi":"10.1016/j.ejrad.2024.111712","DOIUrl":"10.1016/j.ejrad.2024.111712","url":null,"abstract":"<div><h3>Background</h3><p>Brain metastases (BMs) represents a severe neurological complication stemming from cancers originating from various sources. It is a highly challenging clinical task to accurately distinguish the pathological subtypes of brain metastatic tumors from lung cancer (LC).The utility of 2.5-dimensional (2.5D) deep learning (DL) in distinguishing pathological subtypes of LC with BMs is yet to be determined.</p></div><div><h3>Methods</h3><p>A total of 250 patients were included in this retrospective study, divided in a 7:3 ratio into training set (N=175) and testing set (N=75). We devised a method to assemble a series of two-dimensional (2D) images by extracting adjacent slices from a central slice in both superior-inferior and anterior-posterior directions to form a 2.5D dataset. Multi-Instance learning (MIL) is a weakly supervised learning method that organizes training instances into “bags” and provides labels for entire bags, with the purpose of learning a classifier based on the labeled positive and negative bags to predict the corresponding class for an unknown bag. Therefore, we employed MIL to construct a comprehensive 2.5D feature set. Then we used the single-slice as input for constructing the 2D model. DL features were extracted from these slices using the pre-trained ResNet101. All feature sets were inputted into the support vector machine (SVM) for evaluation. The diagnostic performance of the classification models were evaluated using five-fold cross-validation, with accuracy and area under the curve (AUC) metrics calculated for analysis.</p></div><div><h3>Results</h3><p>The optimal performance was obtained using the 2.5D DL model, which achieved the micro-AUC of 0.868 (95% confidence interval [CI], 0.817–0.919) and accuracy of 0.836 in the test cohort. The 2D model achieved the micro-AUC of 0.836 (95 % CI, 0.778–0.894) and accuracy of 0.827 in the test cohort.</p></div><div><h3>Conclusions</h3><p>The proposed 2.5D DL model is feasible and effective in identifying pathological subtypes of BMs from lung cancer.</p></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":null,"pages":null},"PeriodicalIF":3.2,"publicationDate":"2024-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142117728","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}