Quantitative Imaging in Medicine and Surgery最新文献

{"title":"Evaluation of myocardial structure and function in hypertrophic obstructive cardiomyopathy via cardiovascular magnetic resonance: regional distribution and sex differences.","authors":"Yuwei Bao, Peijun Zhao, Lu Huang, Lingping Ran, Dazhong Tang, Xianghu Yan, Liming Xia","doi":"10.21037/qims-2025-457","DOIUrl":"10.21037/qims-2025-457","url":null,"abstract":"<p><strong>Background: </strong>Although hypertrophy and fibrosis have been identified as contributors to mechanical dysfunction in patients with hypertrophic cardiomyopathy (HCM), their regional distribution and sex differences remain inadequately characterized in those with hypertrophic obstructive cardiomyopathy (HOCM). This study aimed to evaluate the determinants of myocardial strain deterioration across different regions and sexes of patients with HOCM through use of cardiovascular magnetic resonance feature tracking (CMR-FT) techniques.</p><p><strong>Methods: </strong>This retrospective study included 102 patients with HOCM who underwent 3.0-T CMR. Global (G) and reginal left ventricular (LV) myocardial peak-strain parameters, including global longitudinal strain (GLS), global circumferential strain (GCS), and global radial strain (GRS), were derived via CMR-FT. Structural metrics including LV wall thickness (WT), hypertrophic distribution, late gadolinium enhancement percentage (LGE%), and extracellular volume fraction (ECV%) were comprehensively assessed. Bivariate correlation and multivariate linear regression analyses were performed to evaluate the independent associations and interaction effects of hypertrophic regions and sex on strain indices.</p><p><strong>Results: </strong>Strain indices exhibited stronger correlations with hypertrophy as compared to fibrosis markers, with correlation coefficients of GLS, GCS, and GRS of -0.72, -0.58, and 0.55, respectively, while mean WT had stronger correlations than did LGE%, at -0.56, -0.55, and 0.51, respectively. Mean WT demonstrated the strongest correlation with GLS, outperforming max WT, hypertrophic segment count, and WT variation. Mean WT emerged as the strongest independent predictor of global strain (GLS: B=0.79; GCS: B=0.54; GRS: B=-1.38; all P values <0.001). ECV% was also independently associated with GLS (B=0.25) and GCS (B=0.16), while WT variation was associated with GCS (B=5.49) (all P values <0.05). Interaction analyses revealed significantly greater strain impairment in female patients and apical regions as WT increased (all P values <0.001).</p><p><strong>Conclusions: </strong>Mean WT emerged as the most robust structural determinant of myocardial strain impairment in patients with HOCM, surpassing other hypertrophic characteristics and fibrosis surrogates. Notably, functional deterioration was most prominent in the apical regions and among female patients. These findings highlight the importance of incorporating sex-specific and region-specific hypertrophic patterns into individualized diagnostic and therapeutic strategies to improve clinical outcomes.</p>","PeriodicalId":54267,"journal":{"name":"Quantitative Imaging in Medicine and Surgery","volume":"15 9","pages":"8594-8610"},"PeriodicalIF":2.3,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12397711/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144978594","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"HDTV: a high-order directional total variation reconstruction algorithm from sparse and limited-angle data in electron paramagnetic resonance imaging.","authors":"Yanjun Zhang, Peng Liu, Chenyun Fang, Yarui Xi, Zhiwei Qiao","doi":"10.21037/qims-2025-8","DOIUrl":"10.21037/qims-2025-8","url":null,"abstract":"<p><strong>Background: </strong>Electron paramagnetic resonance imaging (EPRI)-based oxygen imaging technology enables adaptive radiation therapy, thereby improving tumor control rates. However, the long scanning time limits the development of EPRI. In this study, we endeavored to reduce the scanning time. The general method is sparse reconstruction; if it can be collected in limited-angle range under sparse conditions, the scanning time can be further shortened.</p><p><strong>Methods: </strong>Based on the abovementioned theory, we performed sparse acquisition based on limited-angle range to further accelerate scanning. Moreover, high-order constraints were introduced into the directional total variation (DTV) algorithm to suppress staircase artifacts, and we proposed a high-order DTV (HDTV) model and derived the Chambolle-Pock (CP) solving algorithm. We aimed to realize three-dimensional (3D) sparse and limited-angle EPRI with high precision and thus accelerate the scanning time.</p><p><strong>Results: </strong>The correctness of the HDTV-CP algorithm was validated on simulation data and the limited-angle and sparse reconstruction ability was investigated using real data. The results indicate that the HDTV method effectively suppresses limited-angle artifacts, sparse artifacts, and staircase artifacts while preserving the edge and texture features. Our method showed significant improvements compared to the classic TV method. The normalized root mean square error (nRMSE) decreased from 0.34 to 0.16, and the Pearson correlation coefficient (PCC) increased from 0.93 to 0.98 based on 50 views within half the angular range.</p><p><strong>Conclusions: </strong>For the first time, we combined the limited-angle and sparse problems. The HDTV method may realize 16 times acceleration while ensuring the imaging quality in certain situations. The findings of this study can also extend to the field of limited-angle computed tomography (CT) image reconstruction.</p>","PeriodicalId":54267,"journal":{"name":"Quantitative Imaging in Medicine and Surgery","volume":"15 9","pages":"8471-8490"},"PeriodicalIF":2.3,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12397748/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144978688","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"YOLOv8-BCD: a real-time deep learning framework for pulmonary nodule detection in computed tomography imaging.","authors":"Wenjun Zhu, Xinyue Wang, Jie Xing, Xu Steven Xu, Min Yuan","doi":"10.21037/qims-2025-824","DOIUrl":"10.21037/qims-2025-824","url":null,"abstract":"<p><strong>Background: </strong>Lung cancer remains one of the malignant tumors with the highest global morbidity and mortality rates. Detecting pulmonary nodules in computed tomography (CT) images is essential for early lung cancer screening. However, traditional detection methods often suffer from low accuracy and efficiency, limiting their clinical effectiveness. This study aims to devise an advanced deep-learning framework capable of achieving high-precision, rapid identification of pulmonary nodules in CT imaging, thereby facilitating earlier and more accurate diagnosis of lung cancer.</p><p><strong>Methods: </strong>To address these issues, this paper proposes an improved deep-learning framework named YOLOv8-BCD, based on YOLOv8 and integrating the BiFormer attention mechanism, Content-Aware ReAssembly of Features (CARAFE) up-sampling method, and Depth-wise Over-Parameterized Depth-wise Convolution (DO-DConv) enhanced convolution. To overcome common challenges such as low resolution, noise, and artifacts in lung CT images, the model employs Super-Resolution Generative Adversarial Network (SRGAN)-based image enhancement during preprocessing. The BiFormer attention mechanism is introduced into the backbone to enhance feature extraction capabilities, particularly for small nodules, while CARAFE and DO-DConv modules are incorporated into the head to optimize feature fusion efficiency and reduce computational complexity.</p><p><strong>Results: </strong>Experimental comparisons using 550 CT images from the LUng Nodule Analysis 2016 dataset (LUNA16 dataset) demonstrated that the proposed YOLOv8-BCD achieved detection accuracy and mean average precision (mAP) at an intersection over union (IoU) threshold of 0.5 (mAP_0.5) of 86.4% and 88.3%, respectively, surpassing YOLOv8 by 2.2% in accuracy, 4.5% in mAP_0.5. Additional evaluation on the external TianChi lung nodule dataset further confirmed the model's generalization capability, achieving an mAP_0.5 of 83.8% and mAP_0.5-0.95 of 43.9% with an inference speed of 98 frames per second (FPS).</p><p><strong>Conclusions: </strong>The YOLOv8-BCD model effectively assists clinicians by significantly reducing interpretation time, improving diagnostic accuracy, and minimizing the risk of missed diagnoses, thereby enhancing patient outcomes.</p>","PeriodicalId":54267,"journal":{"name":"Quantitative Imaging in Medicine and Surgery","volume":"15 9","pages":"8189-8204"},"PeriodicalIF":2.3,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12397618/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144978697","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Non-rigid image-volume registration for human livers in laparoscopic surgery.","authors":"Zhenggang Cao, Le Xie, Yuchen Yang","doi":"10.21037/qims-2025-387","DOIUrl":"10.21037/qims-2025-387","url":null,"abstract":"<p><strong>Background: </strong>The fusion of intraoperative 2D laparoscopic images with preoperative 3D scans offers significant advantages in minimally invasive surgery, such as improved spatial understanding and enhanced navigation. This study aims to enable augmented reality for deformable organs through accurate 2D-3D registration. However, achieving real-time and precise alignment remains a major challenge due to organ deformation, occlusion, and the difficulty of estimating camera parameters from monocular images.</p><p><strong>Methods: </strong>We introduce a non-rigid image-volume registration (NRIVR) framework designed specifically for deformable human organs. Our approach employs a long short-term memory-based camera estimation neural network (LCENN) to predict camera poses directly from 2D anatomical contours extracted from laparoscopic images. By leveraging a differentiable mapping from 2D boundaries to camera parameters, the system enables real-time inference. Non-rigid registration is then performed in 2D space by integrating both the projected mesh and estimated deformation fields, ensuring consistent alignment across views.</p><p><strong>Results: </strong>Our experiments, evaluating the contour mapping neural network on laparoscopic images from cholecystectomy, showed that using an LCENN can efficiently predict the camera pose from 2D boundaries, achieving a minimal rotational error of 0.35±0.44° and a translational error of 0.51±0.31 mm. Consequently, our proposed framework effectively achieved 2D-3D registration on a clinical dataset, with an average target registration error of 2.74±1.51 mm.</p><p><strong>Conclusions: </strong>These results validate the feasibility and effectiveness of the proposed method for real-time 2D-3D registration in laparoscopic surgery, paving the way for enhanced image guidance in clinical workflows.</p>","PeriodicalId":54267,"journal":{"name":"Quantitative Imaging in Medicine and Surgery","volume":"15 9","pages":"8440-8456"},"PeriodicalIF":2.3,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12397628/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144978704","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Transcatheter arterial embolization for retroperitoneal bleeding in patients with lumbar artery injury and essential thrombocythemia: a case description.","authors":"Yang An, Huixin Ye","doi":"10.21037/qims-24-2435","DOIUrl":"10.21037/qims-24-2435","url":null,"abstract":"","PeriodicalId":54267,"journal":{"name":"Quantitative Imaging in Medicine and Surgery","volume":"15 9","pages":"8722-8727"},"PeriodicalIF":2.3,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12397691/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144978721","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Late-onset familial cerebral cavernous malformation without a family history: a case description.","authors":"Zhuangzhuang Zhang, Weiping Sun, Zhaoxia Wang, Luhua Wei","doi":"10.21037/qims-2025-760","DOIUrl":"10.21037/qims-2025-760","url":null,"abstract":"","PeriodicalId":54267,"journal":{"name":"Quantitative Imaging in Medicine and Surgery","volume":"15 9","pages":"8717-8721"},"PeriodicalIF":2.3,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12397702/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144978730","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Prognostic value of left atrial strain analysis using cardiac magnetic resonance feature tracking technology for patients with acute myocarditis.","authors":"Yu Chen, Chaohui Tang, Guokun Wang, Yunling Li, Yanming Zhao, Jianxiu Lian, Bing Xu, Changbin Sun, Zhenhua Li, Shuo Li, Cheng Jin, Bo Yu, Shengliang Liu, Xia Gu","doi":"10.21037/qims-2025-849","DOIUrl":"10.21037/qims-2025-849","url":null,"abstract":"<p><strong>Background: </strong>Acute myocarditis (AM) is induced by multiple factors, and its clinical symptoms and prognosis vary significantly. Existing diagnostic and prognostic assessment methods have certain limitations. Myocardial strain analysis based on cardiac magnetic resonance feature tracking (CMR-FT) can detect myocardial damage in the subclinical stage, but studies on the correlation between left atrial (LA) strain and the prognosis of AM are relatively limited. This study aimed to assess whether impaired LA strain parameters based on CMR-FT-derived techniques can be used to predict AM outcomes.</p><p><strong>Methods: </strong>A retrospective cohort study was conducted on the clinical and functional parameters, including LA strain, of 91 patients diagnosed with AM between September 2020 and June 2022. The patients were stratified into the following two groups according to the presence or absence of major adverse cardiovascular events (MACEs): the MACE group, and the non-MACE group. The MACEs identified in our study were: (I) cardiovascular death; (II) hospital admission for heart failure; (III) severe arrhythmia (frequent ventricular extrasystole, ventricular tachycardia, or ventricular fibrillation); and (IV) dilated cardiomyopathy (DCM). The receiver operating characteristic (ROC) curve was used to define a threshold value for the classification between MACEs and no MACEs, and a Cox regression analysis was then used to evaluate the effects of the cardiac magnetic resonance (CMR)-derived parameters on the occurrence of MACEs in the AM patients.</p><p><strong>Results: </strong>Of the 91 AM patients (34.9±18.6 years; 51 males), 13 (14.3%) experienced MACEs during the 1-year follow-up period. In our study, LA longitudinal strain parameters, including reservoir, conduit, and booster function, were significantly reduced in the AM patients in the MACE group compared to the non-MACE group. Specifically, LA reservoir strain was 44.4% (interquartile range, 31.1-66.1%) and 20.6% (interquartile range, 9.9-33.1%) (P=0.002), conduit strain was 22.6% (interquartile range, 15.8-35.1%) and 10.5% (interquartile range, 4.7-13.0%) (P=0.003), and booster strain was 21.4%±14.7% and 10.6%±7.7% (P=0.004) in the non-MACE group and the MACE group, respectively. Among the parameters, reservoir LA longitudinal strain was valuable in diagnosing the development of MACEs in the AM patients [area under the curve (AUC) =0.772, P<0.05], and the survival rate was higher in those with reservoir LA longitudinal strain >25.6%.</p><p><strong>Conclusions: </strong>CMR-derived LA reservoir strain could be used to predict the long-term survival prognosis of AM patients.</p>","PeriodicalId":54267,"journal":{"name":"Quantitative Imaging in Medicine and Surgery","volume":"15 9","pages":"7862-7873"},"PeriodicalIF":2.3,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12397619/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144978732","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Abnormal structural changes and disturbed functional connectivity in patients with Crohn's disease and abdominal pain: a voxel-based morphometry and functional magnetic resonance imaging study.","authors":"Xinyan Wu, Ling Yang, Li Yu, Lingqin Zhang, Nian Liu, Xiaojun Lu, Kang Li","doi":"10.21037/qims-2024-2572","DOIUrl":"10.21037/qims-2024-2572","url":null,"abstract":"<p><strong>Background: </strong>Abdominal pain is a prevalent and debilitating manifestation of Crohn's disease (CD) that significantly impacts the lives of those affected. The neurological pathways responsible for abdominal pain in patients with CD remain unidentified. Therefore, the purpose of this study was to characterize the structural alterations in the brain and associated functional connectivity (FC) in patients with CD and abdominal pain.</p><p><strong>Methods: </strong>The data for three-dimensional T1-weighted and resting-state functional magnetic resonance imaging (fMRI) were gathered from 23 patients with CD and abdominal pain (pain CD), 24 patients with CD but without abdominal pain (nonpain CD), and 25 healthy controls (HCs). Differences in gray-matter volume (GMV) and FC between the pain CD group, nonpain CD group, and HCs were evaluated via analysis of covariance. Biased correlation analyses were employed to evaluate the association of variations in GMV and FC with clinical measures.</p><p><strong>Results: </strong>Voxel-based morphometry analysis revealed that the pain CD group exhibited changes in GMV in the right anterior cingulate cortex (ACC) and orbitofrontal regions, including the orbital parts of the superior frontal gyri, middle frontal gyri (ORBmid), and inferior frontal gyri, as compared to both the HC and nonpain CD groups. Additionally, compared to the HC group, the nonpain CD group showed increased GMV in the bilateral hippocampus. FC analysis showed that the pain CD group had enhanced FC between the right ACC and the default mode network (DMN), particularly with the parahippocampal gyrus (PHG), Rolandic operculum, and postcentral gyrus, as compared to the nonpain CD group. Furthermore, compared to both the nonpain CD and HC groups, pain CD group exhibited increased FC between the left ORBmid and key pain-processing hubs, including the left thalamus, left ACC, and right middle frontal gyrus (MFG). Notably, the FC between the ACC and PHG was negatively correlated with Beck Depression Inventory score (r=-0.548; P=0.019). The FC between the left ORBmid and the right MFG showed a significant negative correlation with Pain Sensitivity Questionnaire score (r=-0.495; P=0.037).</p><p><strong>Conclusions: </strong>Our results suggest that pain may differentially affect brain morphology and function in patients with CD, particularly involving the ACC and orbitofrontal cortex. Specifically, increased FC between the ACC and DMN, as well as orbitofrontal-thalamic circuits, provide novel imaging evidence for the neural mechanisms underlying visceral pain in CD.</p>","PeriodicalId":54267,"journal":{"name":"Quantitative Imaging in Medicine and Surgery","volume":"15 9","pages":"8265-8281"},"PeriodicalIF":2.3,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12397677/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144978390","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Drug-coated balloon with bailout stenting versus drug-eluting stent plus drug-coated balloon in TransAtlantic Inter-Society Consensus C and D femoropopliteal lesions: a propensity score-matched analysis.","authors":"Caibo Chen, Haitao Guan, Siyuan Shen, Pengyu Li, Kang She, Gong Cheng, Guochen Niu, Ziguang Yan, Ziping Yao, Yinghua Zou, Min Yang, Bihui Zhang","doi":"10.21037/qims-2025-367","DOIUrl":"10.21037/qims-2025-367","url":null,"abstract":"<p><strong>Background: </strong>Drug-delivering devices have shown efficacy in clinical trials and are widely used for femoropopliteal artery disease. However, the optimal strategy for complex lesions, such as TransAtlantic Inter-Society Consensus (TASC) C and D lesions, remains debated in real-world practice. This propensity score-matched study aimed to compare the mid-term outcomes between a double-drug strategy [drug-coated balloon (DCB) combined with systemic drug-eluting stents (DES)] and a DCB bailout strategy (DCB with bailout bare-metal stents) in patients with TASC C and D femoropopliteal lesions.</p><p><strong>Methods: </strong>This retrospective single-center study included TASC C and D femoropopliteal patients treated with DCB from October 2016 to July 2024. Propensity score matching (PSM) was performed in a 1:3 ratio, with one patient in the double-drug strategy group for every three in the DCB bailout group. The primary endpoint was 24-month primary patency. Secondary endpoints included freedom from clinically-driven target lesion revascularization (CD-TLR), mortality, complications, symptom improvement, and risk factors for restenosis.</p><p><strong>Results: </strong>After PSM, 32 pairs of patients were analyzed. Baseline characteristics were well-balanced [standardized mean difference (SMD) <0.2 for all covariates]. Primary patency rates at 24 months were comparable (double-drug <i>vs.</i> DCB bailout: 64.5% <i>vs.</i> 76.4%, P=0.76). Freedom from CD-TLR showed no significant difference at 24 months (double-drug <i>vs.</i> DCB bailout: 95.8% <i>vs.</i> 79.1%, P=0.20). The double drug group demonstrated superior Rutherford category improvement (P=0.042). Mortality and complication rates were similar between groups. Dyslipidemia was identified as an independent predictor of loss of primary patency [hazard ratio (HR) =3.03, P=0.024].</p><p><strong>Conclusions: </strong>The double-drug strategy and DCB bailout strategy yielded comparable 24-month patency and freedom from target lesion revascularization (TLR) in TASC C and D lesions.</p>","PeriodicalId":54267,"journal":{"name":"Quantitative Imaging in Medicine and Surgery","volume":"15 9","pages":"7727-7737"},"PeriodicalIF":2.3,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12397624/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144978413","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Combining curriculum learning and weakly supervised attention for enhanced thyroid nodule assessment in ultrasound imaging.","authors":"Chadaporn Keatmanee, Dittapong Songsaeng, Songphon Klabwong, Yoichi Nakaguro, Alisa Kunapinun, Mongkol Ekpanyapong, Matthew N Dailey","doi":"10.21037/qims-24-2431","DOIUrl":"10.21037/qims-24-2431","url":null,"abstract":"<p><strong>Background: </strong>The accurate assessment of thyroid nodules, which are increasingly common with age and lifestyle factors, is essential for early malignancy detection. Ultrasound imaging, the primary diagnostic tool for this purpose, holds promise when paired with deep learning. However, challenges persist with small datasets, where conventional data augmentation can introduce noise and obscure essential diagnostic features. To address dataset imbalance and enhance model generalization, this study integrates curriculum learning with a weakly supervised attention network to improve diagnostic accuracy for thyroid nodule classification.</p><p><strong>Methods: </strong>This study integrates curriculum learning with attention-guided data augmentation to improve deep learning model performance in classifying thyroid nodules. Using verified datasets from Siriraj Hospital, the model was trained progressively, beginning with simpler images and gradually incorporating more complex cases. This structured learning approach is designed to enhance the model's diagnostic accuracy by refining its ability to distinguish benign from malignant nodules.</p><p><strong>Results: </strong>Among the curriculum learning schemes tested, schematic IV achieved the best results, with a precision of 100% for benign and 70% for malignant nodules, a recall of 82% for benign and 100% for malignant, and F1-scores of 90% and 83%, respectively. This structured approach improved the model's diagnostic sensitivity and robustness.</p><p><strong>Conclusions: </strong>These findings suggest that automated thyroid nodule assessment, supported by curriculum learning, has the potential to complement radiologists in clinical practice, enhancing diagnostic accuracy and aiding in more reliable malignancy detection.</p>","PeriodicalId":54267,"journal":{"name":"Quantitative Imaging in Medicine and Surgery","volume":"15 9","pages":"8579-8593"},"PeriodicalIF":2.3,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12397710/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144978643","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}