Polish journal of radiology最新文献

{"title":"On patient-level splitting, contrast-free claims and unsupported comparators in \"Machine learning-based classification of multiple sclerosis lesion activity using multi-sequence MRI radiomics\".","authors":"Stefania Galassi","doi":"10.5114/pjr/213873","DOIUrl":"10.5114/pjr/213873","url":null,"abstract":"","PeriodicalId":94174,"journal":{"name":"Polish journal of radiology","volume":"91 ","pages":"e56-e57"},"PeriodicalIF":0.0,"publicationDate":"2026-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13169490/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147944345","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An automated tuberculosis detection approach using deep learning and machine learning techniques from chest X-ray images: a step towards effective diagnosis.","authors":"Abu Saleh Mohammad Nabil, Md Musfiqur Rahman Mridha, Md Shahid Ahammed Shakil, Iffath Tanjim Moon, Md Fayzul Islam, Md Jamil Chaudhary, Sakib Imtiaz, Md Taufiq Khan","doi":"10.5114/pjr/213688","DOIUrl":"10.5114/pjr/213688","url":null,"abstract":"<p><strong>Purpose: </strong>Tuberculosis (TB) is a severe bacterial infectious lung disease. Millions of people die or experience severe health complications due to TB each year. Accurate, automated, and effective detection of TB is key to curing and preventing person-to-person transmission. In this regard, deep learning (DL) and machine learning (ML) techniques applied to chest X-ray (CXR) images have proved effective. Here, we present our DL- and ML-based approach for TB detection using CXR images.</p><p><strong>Material and methods: </strong>We implemented convolutional neural network (CNN)-based pre-trained DL models, such as DenseNet121, DenseNet169, DenseNet201, ResNet152, and VGG19, as feature extractors. ML models, including support vector machine (SVM), XGBoost, logistic regression, and a DL-based custom model, were used as classifiers. A total of 2,391 CXR images from three publicly available datasets were considered.</p><p><strong>Results: </strong>We found three models achieving the highest value in different evaluation metrics: accuracy (99.91%) with ResNet152 and SVM; recall (99.22%), precision (99.23%), and F1-score (99.22%) with DenseNet169 and the custom classifier; and area under the curve (99.99%) with DenseNet201 and the custom classifier. Among these models, we propose DenseNet169 with the custom classifier as the best performer for potential clinical application, as it reflected a high and well-balanced performance across all evaluation metrics.</p><p><strong>Conclusions: </strong>This study evaluated pre-trained CNN-based DL and ML models on pre-processed CXR images for the detection of TB. The DenseNet169 model with a custom classifier stands out with its high and well-balanced performance, offering a significant contribution to the effective and automated detection of TB.</p>","PeriodicalId":94174,"journal":{"name":"Polish journal of radiology","volume":"91 ","pages":"e46-e55"},"PeriodicalIF":0.0,"publicationDate":"2026-02-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13169492/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147944381","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Comments on data balance, test-set accounting, and thresholded reporting in an adaptive convolution neural network model for tuberculosis detection and diagnosis using semantic segmentation.","authors":"Emmanuel Pio Pastore","doi":"10.5114/pjr/211474","DOIUrl":"10.5114/pjr/211474","url":null,"abstract":"","PeriodicalId":94174,"journal":{"name":"Polish journal of radiology","volume":"91 ","pages":"e45"},"PeriodicalIF":0.0,"publicationDate":"2026-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12907789/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146215292","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Volumetric differences in the brain in children with hypoxic-ischemic encephalopathy after therapeutic hypothermia: a review of MRI findings.","authors":"Bartłomiej Syzdoł, Kamila Ćwik, Magdalena Woźniak","doi":"10.5114/pjr/211222","DOIUrl":"10.5114/pjr/211222","url":null,"abstract":"<p><p>Hypoxic-ischemic encephalopathy (HIE) remains a leading cause of neonatal brain injury, often resulting in long-term neurodevelopmental impairment. Therapeutic hypothermia (TH) is currently the only evidence-based treatment for HIE. This review summarizes volumetric magnetic resonance imaging (MRI) findings in children with HIE treated with TH. Across multiple studies, the hippocampus, thalamus, cerebellum, and basal ganglia consistently exhibit reduced volumes compared to healthy controls, though the statistical significance of these differences varies. Hippocampal volume reductions are observed in the neonatal period and later childhood, and are associated with poorer memory and cognitive outcomes. Thalamic and basal ganglia volumes are lower from the neonatal period through later childhood. Reduced thalamic volume is associated with impairments in cognitive and motor function. Cerebellar volume findings are inconsistent. Interpretation of volumetric differences remains challenging due to methodological heterogeneity across studies. Future studies are needed to validate volumetric biomarkers and establish normative data, facilitating the integration of volumetric MRI into early diagnostic and therapeutic strategies for children with HIE.</p>","PeriodicalId":94174,"journal":{"name":"Polish journal of radiology","volume":"91 ","pages":"e37-e44"},"PeriodicalIF":0.0,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12907790/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146215273","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Prognostic value of collateral scoring systems for hemorrhagic transformation after endovascular treatment in acute ischemic stroke.","authors":"Tamaz Jaoshvili, Natia Burjanadze, Medea Chichua, Mariam Chikhradze, Nikoloz Sainishvili","doi":"10.5114/pjr/211737","DOIUrl":"10.5114/pjr/211737","url":null,"abstract":"<p><strong>Purpose: </strong>Hemorrhagic transformation (HT) represents a serious complication in the management of acute ischemic stroke (AIS), particularly following endovascular treatment (EVT). While clinical predictors of HT have been widely studied, the prognostic value of imaging-based collateral scoring systems remains less well defined. The main purpose of the study is to evaluate the prognostic accuracy of four commonly used collateral scoring systems - Miteff, Maas, modified Tan, the Alberta Stroke Program Early CT Score - and combinations of them, with established variables for predicting HT in AIS patients treated with EVT beyond the 4.5-hour window.</p><p><strong>Material and methods: </strong>This retrospective single-center study included 162 AIS patients who underwent EVT between 2017 and 2023. Patients with baseline computed tomography angiography and follow-up computed tomography were included; those receiving thrombolysis were excluded. Collateral status was assessed using four scoring systems. HT was confirmed by imaging. Patients were divided into HT and non-HT groups. Statistical analysis included χ² tests, binary logistic regression, and ROC curve analysis and evaluation of predictive models (Collateral Model and Integrated Clinical-Collateral Model). Significance was set at <i>p</i> < 0.05.</p><p><strong>Results: </strong>None of the individual collateral scoring systems reached statistical significance as independent predictors of HT. The Collateral Model achieved an area under the curve (AUC) of 0.837 (95% confidence interval [CI]: 0.769-0.906), indicating acceptable discrimination. The Integrated Clinical-Collateral Model further improved predictive performance, with an AUC of 0.933 (95% CI: 0.883-0.983), reflecting excellent accuracy.</p><p><strong>Conclusions: </strong>While individual collateral scoring systems showed limited value in predicting HT risk, their combined use improved prognostic accuracy. The predictive accuracy was even better when collateral scores were integrated with established clinical and imaging predictors.</p>","PeriodicalId":94174,"journal":{"name":"Polish journal of radiology","volume":"91 ","pages":"e30-e36"},"PeriodicalIF":0.0,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12907791/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146215298","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Impact of breast implant presence and type on mean glandular dose calculations.","authors":"Marcin Syc, Aleksandra Bartnik, Beata Brzozowska, Witold Skrzyński","doi":"10.5114/pjr/211318","DOIUrl":"10.5114/pjr/211318","url":null,"abstract":"<p><strong>Purpose: </strong>Breast cancer remains the most frequently diagnosed cancer among women, making mammography a vital tool for its early detection. However, the increasing prevalence of breast augmentation raises concerns about the accuracy of radiation dose estimation during mammography.</p><p><strong>Material and methods: </strong>This study investigates the impact of breast implants on mean glandular dose (MGD) calculations in mammography using Monte Carlo simulations. The research uses the GATE simulation framework to model different implant types (saline, silicone, and adipose tissue) to evaluate their influence on the conversion factors used to determine MGD.</p><p><strong>Results: </strong>The results indicate that the standard conversion factors, derived for non-augmented breasts, may not accurately reflect MGD in augmented breasts.</p><p><strong>Conclusions: </strong>The study highlights the need for tailored conversion factors to ensure precise radiation dose estimation in women with implants, ultimately improving individualized mammography protocols.</p>","PeriodicalId":94174,"journal":{"name":"Polish journal of radiology","volume":"91 ","pages":"e20-e29"},"PeriodicalIF":0.0,"publicationDate":"2026-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12907793/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146215231","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Distribution, extent, and calculated age of initiation of coronary artery calcification.","authors":"Jarosław Piotr Wasilewski, Paweł Raif, Michał Skrzypek, Monika Zofia Czaja-Ziółkowska, Leszek Jan Kukulski, Mariusz Gąsior","doi":"10.5114/pjr/210491","DOIUrl":"10.5114/pjr/210491","url":null,"abstract":"<p><strong>Purpose: </strong>The aim of this study was to characterize the distribution of coronary plaque burden and gender differences in the estimated age of coronary artery calcium (CAC) initiation in patients with moderate or low probability of coronary artery disease using cardiac computed tomography imaging.</p><p><strong>Material and methods: </strong>CAC was measured using the Agatston method. The patients were categorized into risk categories based on CAC score (CACs) and based on the number of involved coronary vessels. We investigated CAC distribution and plaque burden throughout the coronary vasculature in both sexes. Furthermore, we calculated the calcification initiation in each main coronary artery territory, assuming that average annual CACs progression is 25% per year.</p><p><strong>Results: </strong>Among the 937 affected vessels, calcifications were most prevalent in the left anterior descending artery (LAD), with a higher plaque burden ratio. Among patients with a low CAC burden, the LAD was the most frequently involved vessel. Analysis of the estimated age of CAC initiation reveals that calcification begins to develop earlier in the LAD than in other coronary territories in both sexes. Involvement of the second coronary artery territory after one-vessel disease can be expected after a median of 4.5 years (range 2.2-8), with no significant difference between men and women.</p><p><strong>Conclusions: </strong>Calcified plaque was most commonly identified in the LAD. CAC occurs earlier in the LAD than in other coronary arteries. The particular susceptibility to calcification of LAD is not related to gender.</p>","PeriodicalId":94174,"journal":{"name":"Polish journal of radiology","volume":"91 ","pages":"e11-e19"},"PeriodicalIF":0.0,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12907792/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146215275","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Chronic liver disease detection using deep convolutional neural networks with MRI data: a deep learning approach.","authors":"Elif Zoroğlu Altınkaya, Emre Altınkaya, Emre Emekli, Elif Gündoğdu","doi":"10.5114/pjr/210254","DOIUrl":"10.5114/pjr/210254","url":null,"abstract":"<p><strong>Purpose: </strong>Chronic liver disease (CLD) is a significant health issue, and detection is crucial for effective treatment. This study aimed to develop a deep learning based convolutional neural network (DeepCNN) to differentiate CLD from non-CLD patients using magnetic resonance imaging (MRI) images without segmentation, enhancing diagnostic accuracy and supporting timely intervention.</p><p><strong>Material and methods: </strong>A retrospective study was conducted using MRI data from 184 patients collected between 2018 and 2024, totaling 1112 images (460 normal, 652 CLD). Various MRI sequences, including axial T1, T2, and coronal, were used. The images were preprocessed with resizing, augmentation, and normalization techniques. The DeepCNN model was trained and compared against traditional machine learning (ML) algorithms, including logistic regression, <i>k</i>-nearest neighbor, support vector machines, and random forest. Model performance was evaluated using accuracy, precision, recall, F1-score, and confusion matrices.</p><p><strong>Results: </strong>The DeepCNN model achieved a 93% accuracy and an F1-score of 0.939. Precision and recall for CLD classification were 97% and 98%, respectively. In comparison, traditional ML algorithms performed with accuracies ranging from 72.31% to 83.16%, with random forest achieving the highest. The DeepCNN model significantly outperformed these methods, demonstrating its strength in medical image classification. Using axial-only images reduced accuracy to 86%, showing that coronal views contribute valuable information. Limitation of data constrained learning.</p><p><strong>Conclusions: </strong>The DeepCNN model provides superior accuracy in diagnosing CLD compared to traditional ML methods, using MRI images without segmentation. This approach offers a practical solution for improving CLD detection and paves the way for future enhancements using attention mechanisms and advanced deep learning architectures.</p>","PeriodicalId":94174,"journal":{"name":"Polish journal of radiology","volume":"91 ","pages":"e1-e10"},"PeriodicalIF":0.0,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12907794/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146215252","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Subsequent primary and secondary neoplasms in childhood cancer survivors.","authors":"Monika Bekiesińska-Figatowska, Katarzyna Bilska, Sylwia Szkudlińska-Pawlak, Teresa Klepacka, Anna Raciborska, Irena Koch, Elżbieta Michalak","doi":"10.5114/pjr/213904","DOIUrl":"10.5114/pjr/213904","url":null,"abstract":"<p><strong>Purpose: </strong>To count and review cases of subsequent malignant neoplasms (SMNs) in childhood cancer survivors, which are estimated to occur at a tenfold higher frequency than in the general population.</p><p><strong>Material and methods: </strong>Data of 60 patients with SMNs from a tertiary referral center for pediatric solid tumors were analyzed: gender, age at diagnosis, treatment, time between diagnoses, subsequent cancer's association with radiotherapy, observation time, status: alive/dead, genetic background.</p><p><strong>Results: </strong>Median age at time of diagnosis of the index tumor (IT) was 6.8 years (0.1-22.1), of subsequent neoplasm 14.9 years (2.1-36.6). Median time between diagnosis of first and subsequent neoplasm was 6.3 years (0.8-26.2). In 16 patients, secondary tumors developed in irradiated sites. Eleven patients (18.3%) had cancer predisposition syndromes (CPS): 3 <i>RB1</i> mutation, 5 Li-Fraumeni syndrome, 4 neurofibromatosis type 1; one had both Li-Fraumeni syndrome and <i>RB1</i> mutation. Four patients (3 with CPS) had a third malignant neoplasm, and three had a fourth. Thirty-seven patients (61.6%) are alive, 23 (38.4%) died. Median observation time is 15.0 years (1.3-43.1). The overall 5-year survival rate in the whole group is 85% from diagnosis of IT. The overall 5-year survival rate from diagnosis of the second tumor is 63%.</p><p><strong>Conclusions: </strong>Due to risk of SMNs, life-long cancer screening is required for childhood cancer survivors. In CPS, this risk is multiplied, as it results from both genetic factors and previous treatment; in these patients, multiple primary cancers must be taken into account. When assessing imaging studies of patients with a history of malignancy, clinicians should consider not only recurrence and metastases but also the possibility of a new malignancy of a different histopathological nature.</p>","PeriodicalId":94174,"journal":{"name":"Polish journal of radiology","volume":"90 ","pages":"e621-e631"},"PeriodicalIF":0.0,"publicationDate":"2025-12-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12825110/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146055734","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessing glymphatic dysfunction using the diffusion tensor image analysis along the perivascular space (DTI-ALPS) index in gliomas and metastases.","authors":"Dhanush Jayanna, Shravan Reddy Kankara, Shreyas Reddy K, Arpitha Josephine, Vilas Melrick Tellis, Meghana Kancharla, Aaral Alisha Monteiro, Sunitha Palasamudram Kumaran, Saikanth Reddy Deepalam","doi":"10.5114/pjr/209785","DOIUrl":"10.5114/pjr/209785","url":null,"abstract":"<p><strong>Purpose: </strong>To evaluate glymphatic dysfunction using the diffusion tensor image analysis along the perivascular space (DTI-ALPS) index in patients with low-grade gliomas (LGGs), high-grade gliomas (HGGs), and metastases, assess its feasibility as a non-invasive biomarker for tumour differentiation, and examine its relationship with tumourspecific characteristics.</p><p><strong>Material and methods: </strong>This single-centre retrospective study examined patients with LGGs (<i>n</i> = 30), HGGs (<i>n</i> = 30), and metastases (<i>n</i> = 20), calculating the DTI-ALPS index. Tumour volume, peritumoral oedema, tumour volume-to-total brain volume ratio (TV/TBV), and peritumoral oedema-to-total brain volume ratio (PTE/TBV) were obtained using 3D segmentation. The DTI-ALPS index was compared across the 3 tumour groups, and its relationships with tumour-associated oedema, peritumoral oedema, TV/TBV, and PTE/TBV were analysed within and between the groups. Additionally, the DTI-ALPS index was compared between isocitrate dehydrogenase-1 (IDH1) mutant and IDH1 wild-type gliomas.</p><p><strong>Results: </strong>There was a significant difference in the DTI-ALPS index between the 3 groups (<i>p</i> < 0.001), with HGGs having the lowest DTI-ALPS index values, followed by metastases and LGGs. Receiver operating characteristic (ROC) analysis showed that the DTI-ALPS index had excellent sensitivity and specificity for LGGs (> 1.3838) and HGGs (< 1.317). No significant correlation was found between the DTI-ALPS index and tumour volume, TV/TBV, PTE/TBV, or peritumoral oedema. Furthermore, the mean DTI-ALPS index in IDH1 wild-type gliomas (1.23 ± 0.08) was significantly lower than that observed in IDH1 mutant tumours (1.42 ± 0.10; <i>p</i> < 0.001).</p><p><strong>Conclusion: </strong>The DTI-ALPS index provides valuable insights into glymphatic dysfunction in brain tumours. This study underscores its potential as a non-invasive biomarker in differentiating these tumour groups.</p>","PeriodicalId":94174,"journal":{"name":"Polish journal of radiology","volume":"90 ","pages":"e611-e620"},"PeriodicalIF":0.0,"publicationDate":"2025-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12825112/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146055663","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}