Ioana Hutuca,Kristof L Egervari,Doron Merkler,Maria Isabel Vargas
Myxopapillary ependymomas (MPE), classified as grade 2 tumors by the WHO, are rare spinal neoplasms. Despite their slow growth and generally benign nature, MPE have a high recurrence rate and potential for cerebrospinal fluid dissemination. This study aims to identify the MRI characteristics and pathological patterns of MPE and investigate potential correlations between the MRI characteristics and specific histopathological patterns. We assessed 13 patients (7 men; mean age, 45.1 years) with pathologically proven MPE. MRI images were reviewed for tumor location, size, T1 and T2 signal characteristics, contrast enhancement, hemosiderin cap presence, vertebral scalloping, drop metastasis, and prominent intradural flow voids. Four histopathological patterns (microcystic, solid, hemorrhagic, and high hyalin content) were defined and segmented, with surface areas measured and percentages calculated relative to the total tissue surface. Most tumors were in the lumbar region (84.61%), with MRI revealing typical features such as T2 hyperintensity (100%) and contrast enhancement (92.3%). A rare non-enhancing MPE was noted. Large tumors exhibited a microcystic pathology pattern, with two cases with this pattern showing drop metastasis on MRI. Smaller tumors typically presented a solid pathology pattern with homogenous MRI signals. This study underscores the diverse MRI presentations of MPE and suggests a potential link between microcystic patterns in pathology and large MPE with drop metastasis.ABBREVIATIONS: MPE= myxopapillary ependymoma.
{"title":"The many faces of myxopapillary ependynomas.","authors":"Ioana Hutuca,Kristof L Egervari,Doron Merkler,Maria Isabel Vargas","doi":"10.3174/ajnr.a8499","DOIUrl":"https://doi.org/10.3174/ajnr.a8499","url":null,"abstract":"Myxopapillary ependymomas (MPE), classified as grade 2 tumors by the WHO, are rare spinal neoplasms. Despite their slow growth and generally benign nature, MPE have a high recurrence rate and potential for cerebrospinal fluid dissemination. This study aims to identify the MRI characteristics and pathological patterns of MPE and investigate potential correlations between the MRI characteristics and specific histopathological patterns. We assessed 13 patients (7 men; mean age, 45.1 years) with pathologically proven MPE. MRI images were reviewed for tumor location, size, T1 and T2 signal characteristics, contrast enhancement, hemosiderin cap presence, vertebral scalloping, drop metastasis, and prominent intradural flow voids. Four histopathological patterns (microcystic, solid, hemorrhagic, and high hyalin content) were defined and segmented, with surface areas measured and percentages calculated relative to the total tissue surface. Most tumors were in the lumbar region (84.61%), with MRI revealing typical features such as T2 hyperintensity (100%) and contrast enhancement (92.3%). A rare non-enhancing MPE was noted. Large tumors exhibited a microcystic pathology pattern, with two cases with this pattern showing drop metastasis on MRI. Smaller tumors typically presented a solid pathology pattern with homogenous MRI signals. This study underscores the diverse MRI presentations of MPE and suggests a potential link between microcystic patterns in pathology and large MPE with drop metastasis.ABBREVIATIONS: MPE= myxopapillary ependymoma.","PeriodicalId":7875,"journal":{"name":"American Journal of Neuroradiology","volume":null,"pages":null},"PeriodicalIF":3.5,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142268346","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}
Burak Berksu Ozkara,Alexandre Boutet,Bryan A Comstock,Johan Van Goethem,Thierry A G M Huisman,Jeffrey S Ross,Luca Saba,Lubdha M Shah,Max Wintermark,Mauricio Castillo
BACKGROUND AND PURPOSEWe aimed to evaluate GPT-4's ability to write radiology editorials and to compare these with human-written counterparts, thereby determining their real-world applicability for scientific writing.MATERIALS AND METHODSSixteen editorials from eight journals were included. To generate the AI-written editorials, the summary of 16 human-written editorials was fed into GPT-4. Six experienced editors reviewed the articles. First, an unpaired approach was used. The raters were asked to evaluate the content of each article using a 1-5 Likert scale across specified metrics. Then, they determined whether the editorials were written by humans or AI. The articles were then evaluated in pairs to determine which article was generated by AI and which should be published. Finally, the articles were analyzed with an AI detector and for plagiarism.RESULTSThe human-written articles had a median AI probability score of 2.0%, whereas the AI-written articles had 58%. The median similarity score among AI-written articles was 3%. 58% of unpaired articles were correctly classified regarding authorship. Rating accuracy was increased to 70% in the paired setting. AI-written articles received slightly higher scores in most metrics. When stratified by perception, human-written perceived articles were rated higher in most categories. In the paired setting, raters strongly preferred publishing the article they perceived as human-written (82%).CONCLUSIONSGPT-4 can write high-quality articles that iThenticate does not flag as plagiarized, which may go undetected by editors, and that detection tools can detect to a limited extent. Editors showed a positive bias toward human-written articles.ABBREVIATIONSAI = Artificial intelligence; LLM = large language model; SD = standard deviation.
{"title":"Artificial Intelligence-Generated Editorials in Radiology: Can Expert Editors Detect Them?","authors":"Burak Berksu Ozkara,Alexandre Boutet,Bryan A Comstock,Johan Van Goethem,Thierry A G M Huisman,Jeffrey S Ross,Luca Saba,Lubdha M Shah,Max Wintermark,Mauricio Castillo","doi":"10.3174/ajnr.a8505","DOIUrl":"https://doi.org/10.3174/ajnr.a8505","url":null,"abstract":"BACKGROUND AND PURPOSEWe aimed to evaluate GPT-4's ability to write radiology editorials and to compare these with human-written counterparts, thereby determining their real-world applicability for scientific writing.MATERIALS AND METHODSSixteen editorials from eight journals were included. To generate the AI-written editorials, the summary of 16 human-written editorials was fed into GPT-4. Six experienced editors reviewed the articles. First, an unpaired approach was used. The raters were asked to evaluate the content of each article using a 1-5 Likert scale across specified metrics. Then, they determined whether the editorials were written by humans or AI. The articles were then evaluated in pairs to determine which article was generated by AI and which should be published. Finally, the articles were analyzed with an AI detector and for plagiarism.RESULTSThe human-written articles had a median AI probability score of 2.0%, whereas the AI-written articles had 58%. The median similarity score among AI-written articles was 3%. 58% of unpaired articles were correctly classified regarding authorship. Rating accuracy was increased to 70% in the paired setting. AI-written articles received slightly higher scores in most metrics. When stratified by perception, human-written perceived articles were rated higher in most categories. In the paired setting, raters strongly preferred publishing the article they perceived as human-written (82%).CONCLUSIONSGPT-4 can write high-quality articles that iThenticate does not flag as plagiarized, which may go undetected by editors, and that detection tools can detect to a limited extent. Editors showed a positive bias toward human-written articles.ABBREVIATIONSAI = Artificial intelligence; LLM = large language model; SD = standard deviation.","PeriodicalId":7875,"journal":{"name":"American Journal of Neuroradiology","volume":null,"pages":null},"PeriodicalIF":3.5,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142250136","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}
John P Welby,Nicholas M Baumel,Ghazal S Daher,Armine Kocharyan,Christine M Lohse,Girish Bathla,Matthew L Carlson,John I Lane,John C Benson
BACKGROUND AND PURPOSEVestibular schwannomas (VSs) are benign neurogenic tumors commonly associated with progressive unilateral hearing loss, tinnitus, and vestibular symptoms. Growing evidence links signal changes in the VS-adjacent labyrinth with sensorineural hearing loss. This study seeks to quantify the association of labyrinthine signal on post-gadolinium 3D-FLAIR imaging correlates with hearing loss and to evaluate potential longitudinal changes over time.MATERIALS AND METHODSSelected patients were identified from a prospectively maintained VS registry. Mean signal intensity ratios of the bilateral labyrinth and pons were measured on 3D-FLAIR post-gadolinium MRI. Correlations with paired audiometric data including pure tone average (PTA), word recognition score (WRS), and AAO-HNS hearing class within one year were evaluated.RESULTS125 studies obtained from 2015 to 2022 among 66 patients undergoing observational management for sporadic VS were analyzed. Increased signal intensity was noted of the VS-affected labyrinth/contralateral labyrinth (mean ratio 1.56, SD 0.58). Increased signal intensity was associated with increased PTA on both labyrinthine (correlation coefficient [CC] 0.20, p=0.03) and pontine comparisons (CC 0.24, p=0.006), and with decreased WRS on pontine comparisons (CC -0.18, p=0.04). Increased signal intensity was significantly associated with non-serviceable AAO-HNS C/D hearing when intensities were compared to the pons (p=0.01) but not the contralateral labyrinth (p=0.1). Among 44 patients with available follow-up, no statistically significant associations were identified between audiometric data and signal changes over the same interval.CONCLUSIONSIncreased 3D-FLAIR post-gadolinium labyrinthine signal is associated with sensorineural hearing loss; however, its relationship with hearing trajectory remains unclear. Overall findings suggest that while post-gadolinium 3D-FLAIR techniques are sensitive to inner ear involvement associated with VS, the driving mechanism and their temporal relationships with labyrinthine signal intensity and hearing impairment remain unknown.ABBREVIATIONSAAO-HNS =American Academy of Otolaryngology -Head and Neck Surgery. BLB =blood-labyrinth barrier. CPA = cerebellopontine angle. IAC = internal auditory canal. PTA = pure tone average; SIR = signal intensity ratio. VS = vestibular schwannoma. WRS = word recognition score.
{"title":"Vestibular schwannoma-related increased labyrinthine post-gadolinium 3D-FLAIR signal intensity and association with hearing impairment.","authors":"John P Welby,Nicholas M Baumel,Ghazal S Daher,Armine Kocharyan,Christine M Lohse,Girish Bathla,Matthew L Carlson,John I Lane,John C Benson","doi":"10.3174/ajnr.a8498","DOIUrl":"https://doi.org/10.3174/ajnr.a8498","url":null,"abstract":"BACKGROUND AND PURPOSEVestibular schwannomas (VSs) are benign neurogenic tumors commonly associated with progressive unilateral hearing loss, tinnitus, and vestibular symptoms. Growing evidence links signal changes in the VS-adjacent labyrinth with sensorineural hearing loss. This study seeks to quantify the association of labyrinthine signal on post-gadolinium 3D-FLAIR imaging correlates with hearing loss and to evaluate potential longitudinal changes over time.MATERIALS AND METHODSSelected patients were identified from a prospectively maintained VS registry. Mean signal intensity ratios of the bilateral labyrinth and pons were measured on 3D-FLAIR post-gadolinium MRI. Correlations with paired audiometric data including pure tone average (PTA), word recognition score (WRS), and AAO-HNS hearing class within one year were evaluated.RESULTS125 studies obtained from 2015 to 2022 among 66 patients undergoing observational management for sporadic VS were analyzed. Increased signal intensity was noted of the VS-affected labyrinth/contralateral labyrinth (mean ratio 1.56, SD 0.58). Increased signal intensity was associated with increased PTA on both labyrinthine (correlation coefficient [CC] 0.20, p=0.03) and pontine comparisons (CC 0.24, p=0.006), and with decreased WRS on pontine comparisons (CC -0.18, p=0.04). Increased signal intensity was significantly associated with non-serviceable AAO-HNS C/D hearing when intensities were compared to the pons (p=0.01) but not the contralateral labyrinth (p=0.1). Among 44 patients with available follow-up, no statistically significant associations were identified between audiometric data and signal changes over the same interval.CONCLUSIONSIncreased 3D-FLAIR post-gadolinium labyrinthine signal is associated with sensorineural hearing loss; however, its relationship with hearing trajectory remains unclear. Overall findings suggest that while post-gadolinium 3D-FLAIR techniques are sensitive to inner ear involvement associated with VS, the driving mechanism and their temporal relationships with labyrinthine signal intensity and hearing impairment remain unknown.ABBREVIATIONSAAO-HNS =American Academy of Otolaryngology -Head and Neck Surgery. BLB =blood-labyrinth barrier. CPA = cerebellopontine angle. IAC = internal auditory canal. PTA = pure tone average; SIR = signal intensity ratio. VS = vestibular schwannoma. WRS = word recognition score.","PeriodicalId":7875,"journal":{"name":"American Journal of Neuroradiology","volume":null,"pages":null},"PeriodicalIF":3.5,"publicationDate":"2024-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142250138","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}
A Tietze,B Bison,,J Engelhardt,T Fenouil,D Figarella-Branger,E Goebell,J Hakumäki,E Koscielniak,L E Ludlow,D Meyronet,P Nyman,I Øra,J Pesola,T Rauramaa,R E Reddingius,D Samuel,A Sexton-Oates,A Vasiljevic,A K Wefers,J Zamecnik,Dtw Jones,M K Keck,K von Hoff
BACKGROUND AND PURPOSECNS embryonal tumor with PLAGL1/PLAGL2 amplification (ET, PLAGL) is a newly identified, highly malignant pediatric tumor. Systematic MRI descriptions of ET, PLAGL are currently lacking.MATERIALS AND METHODSMRI data from 19 treatment-naïve patients with confirmed ET, PLAGL were analyzed. Evaluation focused on anatomical involvement, tumor localization, MRI signal characteristics, DWI behavior, and the presence of necrosis and hemorrhage. Descriptive statistics (median, interquartile range, percentage) were assessed.RESULTSTen patients had PLAGL1 and nine PLAGL2 amplifications. The solid components of the tumors were often multinodular with heterogeneous enhancement (mild to intermediate in 47% and intermediate to strong in 47% of cases). Non-solid components included cysts in 47% and necrosis in 84% of the cases. The tumors showed heterogeneous T2WI hyper-and isointensity (74%), relatively little diffusion restriction (ADC values < contralateral normal-appearing WM in 36% of cases with available DWI), and tendencies towards hemorrhage/calcification (42%). No reliable distinction was found between PLAGL1-and PLAGL2-amplified tumors or compared to other embryonal CNS tumors.CONCLUSIONSThe study contributes to understanding the imaging characteristics of ET, PLAGL. It underscores the need for collaboration in studying rare pediatric tumors and advocates for the use of harmonized imaging protocols for better characterization.ABBREVIATIONSATRT= atypical teratoid/rhabdoid tumor; ETMR= embryonal tumor with multilayered rosettes; ET, PLAGL= CNS embryonal tumor with PLAGL amplification; EVD= external ventricular drain; IQR: interquartile range; PLAGL1= pleomorphic adenoma gene-like 1; PLAGL2= pleomorphic adenoma gene-like 2; WHO= World Health Organization.
{"title":"CNS Embryonal Tumor with PLAGL Amplification, a New Tumor Type in Children and Adolescents: Insights from a Comprehensive MRI Analysis.","authors":"A Tietze,B Bison,,J Engelhardt,T Fenouil,D Figarella-Branger,E Goebell,J Hakumäki,E Koscielniak,L E Ludlow,D Meyronet,P Nyman,I Øra,J Pesola,T Rauramaa,R E Reddingius,D Samuel,A Sexton-Oates,A Vasiljevic,A K Wefers,J Zamecnik,Dtw Jones,M K Keck,K von Hoff","doi":"10.3174/ajnr.a8496","DOIUrl":"https://doi.org/10.3174/ajnr.a8496","url":null,"abstract":"BACKGROUND AND PURPOSECNS embryonal tumor with PLAGL1/PLAGL2 amplification (ET, PLAGL) is a newly identified, highly malignant pediatric tumor. Systematic MRI descriptions of ET, PLAGL are currently lacking.MATERIALS AND METHODSMRI data from 19 treatment-naïve patients with confirmed ET, PLAGL were analyzed. Evaluation focused on anatomical involvement, tumor localization, MRI signal characteristics, DWI behavior, and the presence of necrosis and hemorrhage. Descriptive statistics (median, interquartile range, percentage) were assessed.RESULTSTen patients had PLAGL1 and nine PLAGL2 amplifications. The solid components of the tumors were often multinodular with heterogeneous enhancement (mild to intermediate in 47% and intermediate to strong in 47% of cases). Non-solid components included cysts in 47% and necrosis in 84% of the cases. The tumors showed heterogeneous T2WI hyper-and isointensity (74%), relatively little diffusion restriction (ADC values < contralateral normal-appearing WM in 36% of cases with available DWI), and tendencies towards hemorrhage/calcification (42%). No reliable distinction was found between PLAGL1-and PLAGL2-amplified tumors or compared to other embryonal CNS tumors.CONCLUSIONSThe study contributes to understanding the imaging characteristics of ET, PLAGL. It underscores the need for collaboration in studying rare pediatric tumors and advocates for the use of harmonized imaging protocols for better characterization.ABBREVIATIONSATRT= atypical teratoid/rhabdoid tumor; ETMR= embryonal tumor with multilayered rosettes; ET, PLAGL= CNS embryonal tumor with PLAGL amplification; EVD= external ventricular drain; IQR: interquartile range; PLAGL1= pleomorphic adenoma gene-like 1; PLAGL2= pleomorphic adenoma gene-like 2; WHO= World Health Organization.","PeriodicalId":7875,"journal":{"name":"American Journal of Neuroradiology","volume":null,"pages":null},"PeriodicalIF":3.5,"publicationDate":"2024-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142250183","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}
Chris Lee,Mihir Khunte,Kyle Tegtmeyer,Seyedmehdi Payabvash,Melissa M Chen,Suresh Mukherji,Dheeraj Gandhi,Ajay Malhotra
BACKGROUND AND PURPOSEPhysician-industry relationships can be useful for driving innovation and technologic progress, though little is known about the scale or impact of industry involvement in neuroradiology. The purpose of this study was to assess the trends and distributions of industry payments to neuroradiologists.MATERIALS AND METHODSNeuroradiologists were identified using a previously-validated method based on Work Relative Value Units and Neiman Imaging Types of Service classification. Data on payments from industry were obtained from the Open Payments database from the Centers for Medicare & Medicaid Services, from 2016 to 2021. Payments were grouped into 7 categories, including consulting fees, education, gifts, medical supplies, research, royalties/ownership, and speaker fees. Descriptive statistics were calculated.RESULTSA total of 3019 neuroradiologists were identified in this study. Between 2016 and 2021, 48% (1440/3019) received at least 1 payment from industry, amounting to a total number of 21,967 payments. Each year, among those receiving payments from industry, each unique neuroradiologist received between a mean of 5.49-7.42 payments and a median of 2 payments, indicating a strong rightward skew to the distribution of payments. Gifts were the most frequent payment type made (60%, 13,285/21,967) but accounted for only 4.1% ($689,859/$17,010,546) of payment value. The greatest aggregate payment value came from speaker fees, which made up 36% ($6,127,484/$17,010,546) of the total payment value. The top 5% highest paid neuroradiologists received 42% (9133/21,967) of payments, which accounted for 84% ($14,284,120/$17,010,546) of the total dollar value. Since the start of the coronavirus 2019 (COVID-19) pandemic, the number of neuroradiologists receiving industry payments decreased from a mean of 671 neuroradiologists per year prepandemic (2016-2019) to 411 in the postpandemic (2020-2021) era (P = .030). The total number of payments to neuroradiologists decreased from 4177 per year prepandemic versus 2631 per year postpandemic (P = .011).CONCLUSIONSIndustry payments to neuroradiologists are highly concentrated among top earners, particularly among the top 5% of payment recipients. The number of payments decreased during the COVID-19 pandemic, though the dollar value of payments was offset by coincidental increases in royalty payments. Further investigation is needed in subsequent years to determine if the postpandemic changes in industry payment trends continue.
{"title":"Distribution and Disparities of Industry Payments to Neuroradiologists.","authors":"Chris Lee,Mihir Khunte,Kyle Tegtmeyer,Seyedmehdi Payabvash,Melissa M Chen,Suresh Mukherji,Dheeraj Gandhi,Ajay Malhotra","doi":"10.3174/ajnr.a8404","DOIUrl":"https://doi.org/10.3174/ajnr.a8404","url":null,"abstract":"BACKGROUND AND PURPOSEPhysician-industry relationships can be useful for driving innovation and technologic progress, though little is known about the scale or impact of industry involvement in neuroradiology. The purpose of this study was to assess the trends and distributions of industry payments to neuroradiologists.MATERIALS AND METHODSNeuroradiologists were identified using a previously-validated method based on Work Relative Value Units and Neiman Imaging Types of Service classification. Data on payments from industry were obtained from the Open Payments database from the Centers for Medicare & Medicaid Services, from 2016 to 2021. Payments were grouped into 7 categories, including consulting fees, education, gifts, medical supplies, research, royalties/ownership, and speaker fees. Descriptive statistics were calculated.RESULTSA total of 3019 neuroradiologists were identified in this study. Between 2016 and 2021, 48% (1440/3019) received at least 1 payment from industry, amounting to a total number of 21,967 payments. Each year, among those receiving payments from industry, each unique neuroradiologist received between a mean of 5.49-7.42 payments and a median of 2 payments, indicating a strong rightward skew to the distribution of payments. Gifts were the most frequent payment type made (60%, 13,285/21,967) but accounted for only 4.1% ($689,859/$17,010,546) of payment value. The greatest aggregate payment value came from speaker fees, which made up 36% ($6,127,484/$17,010,546) of the total payment value. The top 5% highest paid neuroradiologists received 42% (9133/21,967) of payments, which accounted for 84% ($14,284,120/$17,010,546) of the total dollar value. Since the start of the coronavirus 2019 (COVID-19) pandemic, the number of neuroradiologists receiving industry payments decreased from a mean of 671 neuroradiologists per year prepandemic (2016-2019) to 411 in the postpandemic (2020-2021) era (P = .030). The total number of payments to neuroradiologists decreased from 4177 per year prepandemic versus 2631 per year postpandemic (P = .011).CONCLUSIONSIndustry payments to neuroradiologists are highly concentrated among top earners, particularly among the top 5% of payment recipients. The number of payments decreased during the COVID-19 pandemic, though the dollar value of payments was offset by coincidental increases in royalty payments. Further investigation is needed in subsequent years to determine if the postpandemic changes in industry payment trends continue.","PeriodicalId":7875,"journal":{"name":"American Journal of Neuroradiology","volume":null,"pages":null},"PeriodicalIF":3.5,"publicationDate":"2024-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142250185","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}
Mohamed Sobhi Jabal,Marwa A Mohammed,Cody L Nesvick,Hassan Kobeissi,Christopher S Graffeo,Bruce E Pollock,Waleed Brinjikji
BACKGROUND AND PURPOSEStereotactic radiosurgery is a key treatment modality for cerebral AVMs, particularly for small lesions and those located in eloquent brain regions. Predicting obliteration remains challenging due to evolving treatment paradigms and complex AVM presentations. With digital subtraction angiography (DSA) being the gold standard for outcome evaluation, radiomic approaches offer potential for more objective and detailed analysis. We aimed to develop machine learning modeling using DSA quantitative features for post-SRS obliteration prediction.MATERIALS AND METHODSA prospective registry of patients with cerebral AVMs was screened to include patients with digital prestereotactic radiosurgery DSA. Anterior-posterior and lateral views were retrieved and manually segmented. Quantitative features were computed from the lesion ROI. Following feature selection, machine learning models were developed to predict unsuccessful 2-year total obliteration using processed radiomics features in comparison with clinical and radiosurgical features. When we evaluated through area under the receiver operating characteristic curve (AUROC), accuracy, area under the precision-recall curve F1, recall, and precision, the best performing model predictions on the test set were interpreted using the Shapley additive explanations approach.RESULTSDSA images of 100 included patients were retrieved and analyzed. The best-performing clinical radiosurgical model was a gradient boosting classifier with an AUROC of 68% and a recall of 67%. When we used radiomics variables as input, the AdaBoost classifier had the best evaluation metrics with an AUROC of 79% and a recall of 75%. The most important clinico-radiosurgical features, ranked by model contribution, were lesion volume, patient age, treatment dose rate, the presence of seizure at presentation, and prior resection. The most important ranked radiomics features were the following: gray-level size zone matrix, gray-level nonuniformity, kurtosis, sphericity, skewness, and gray-level dependence matrix dependence nonuniformity.CONCLUSIONSThe combination of radiomics with machine learning is a promising approach for predicting cerebral AVM obliteration status following stereotactic radiosurgery. DSA could enhance prognostication of stereotactic radiosurgery-treated AVMs due to its high spatial resolution. Model interpretation is essential for building transparent models and establishing clinically valid radiomic signatures.
{"title":"DSA Quantitative Analysis and Predictive Modeling of Obliteration in Cerebral AVM following Stereotactic Radiosurgery.","authors":"Mohamed Sobhi Jabal,Marwa A Mohammed,Cody L Nesvick,Hassan Kobeissi,Christopher S Graffeo,Bruce E Pollock,Waleed Brinjikji","doi":"10.3174/ajnr.a8351","DOIUrl":"https://doi.org/10.3174/ajnr.a8351","url":null,"abstract":"BACKGROUND AND PURPOSEStereotactic radiosurgery is a key treatment modality for cerebral AVMs, particularly for small lesions and those located in eloquent brain regions. Predicting obliteration remains challenging due to evolving treatment paradigms and complex AVM presentations. With digital subtraction angiography (DSA) being the gold standard for outcome evaluation, radiomic approaches offer potential for more objective and detailed analysis. We aimed to develop machine learning modeling using DSA quantitative features for post-SRS obliteration prediction.MATERIALS AND METHODSA prospective registry of patients with cerebral AVMs was screened to include patients with digital prestereotactic radiosurgery DSA. Anterior-posterior and lateral views were retrieved and manually segmented. Quantitative features were computed from the lesion ROI. Following feature selection, machine learning models were developed to predict unsuccessful 2-year total obliteration using processed radiomics features in comparison with clinical and radiosurgical features. When we evaluated through area under the receiver operating characteristic curve (AUROC), accuracy, area under the precision-recall curve F1, recall, and precision, the best performing model predictions on the test set were interpreted using the Shapley additive explanations approach.RESULTSDSA images of 100 included patients were retrieved and analyzed. The best-performing clinical radiosurgical model was a gradient boosting classifier with an AUROC of 68% and a recall of 67%. When we used radiomics variables as input, the AdaBoost classifier had the best evaluation metrics with an AUROC of 79% and a recall of 75%. The most important clinico-radiosurgical features, ranked by model contribution, were lesion volume, patient age, treatment dose rate, the presence of seizure at presentation, and prior resection. The most important ranked radiomics features were the following: gray-level size zone matrix, gray-level nonuniformity, kurtosis, sphericity, skewness, and gray-level dependence matrix dependence nonuniformity.CONCLUSIONSThe combination of radiomics with machine learning is a promising approach for predicting cerebral AVM obliteration status following stereotactic radiosurgery. DSA could enhance prognostication of stereotactic radiosurgery-treated AVMs due to its high spatial resolution. Model interpretation is essential for building transparent models and establishing clinically valid radiomic signatures.","PeriodicalId":7875,"journal":{"name":"American Journal of Neuroradiology","volume":null,"pages":null},"PeriodicalIF":3.5,"publicationDate":"2024-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142250184","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}
Mechanical thrombectomy devices have potential to injure the vessel during treatment of acute ischemic stroke. The goal of the current work was to tailor in vitro endothelialized silicone models for stent retriever assessment and to evaluate endothelial injury following treatment by various stent retriever designs and sizes. Clinically-relevant neurovascular geometries were first modeled out of silicone, then sterilized, coated with fibronectin, placed in bioreactors, seeded with human endothelial cells, and cultivated under flow. Several sizes of two different commercially available stent retrievers were then deployed in, and retracted through, vessels. Vessels were immediately harvested and stained. Endothelial injury, identified as denudation, was quantified using ImageJ. Results illustrated that endothelial injury ranged from 16-18% in wire/microcatheter-only treated vessels, 37-61% in 1-pass treatments, and 52-70% in 2-pass treatments. Overall this work showcases an in vitro approach for early stage assessment of the extent and location of vascular injury following stent retriever retraction.
{"title":"In vitro assessment of vascular injury following stent retriever retraction in clinically-relevant endothelialized silicone models.","authors":"Isabelle Starr,Harrison Oen,Alyssa McCulloch,Sergey Frenklakh,Ryan Grandfield,Hana Choe,Kristen O'Halloran Cardinal","doi":"10.3174/ajnr.a8495","DOIUrl":"https://doi.org/10.3174/ajnr.a8495","url":null,"abstract":"Mechanical thrombectomy devices have potential to injure the vessel during treatment of acute ischemic stroke. The goal of the current work was to tailor in vitro endothelialized silicone models for stent retriever assessment and to evaluate endothelial injury following treatment by various stent retriever designs and sizes. Clinically-relevant neurovascular geometries were first modeled out of silicone, then sterilized, coated with fibronectin, placed in bioreactors, seeded with human endothelial cells, and cultivated under flow. Several sizes of two different commercially available stent retrievers were then deployed in, and retracted through, vessels. Vessels were immediately harvested and stained. Endothelial injury, identified as denudation, was quantified using ImageJ. Results illustrated that endothelial injury ranged from 16-18% in wire/microcatheter-only treated vessels, 37-61% in 1-pass treatments, and 52-70% in 2-pass treatments. Overall this work showcases an in vitro approach for early stage assessment of the extent and location of vascular injury following stent retriever retraction.","PeriodicalId":7875,"journal":{"name":"American Journal of Neuroradiology","volume":null,"pages":null},"PeriodicalIF":3.5,"publicationDate":"2024-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142196344","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}
Angela Ayobi,Adam Davis,Peter D Chang,Daniel S Chow,Kambiz Nael,Maxime Tassy,Sarah Quenet,Sylvain Fogola,Peter Shabe,David Fussell,Christophe Avare,Yasmina Chaibi
BACKGROUND AND PURPOSEASPECTS is a long-standing and well documented selection criteria for acute ischemic stroke treatment, however, the interpretation of ASPECTS is a challenging and time-consuming task for physicians with significant interobserver variabilities. We conducted a multi-reader, multi-case study in which readers assessed ASPECTS without and with the support of a deep learning (DL)-based algorithm in order to analyze the impact of the software on clinicians' performance and interpretation time.MATERIALS AND METHODSA total of 200 NCCT scans from 5 clinical sites (27 scanner models, 4 different vendors) were retrospectively collected. Reference standard was established through the consensus of three expert neuroradiologists who had access to baseline CTA and CTP data. Subsequently, eight additional clinicians (four typical ASPECTS reader and four senior neuroradiologists) analyzed the NCCT scans without and with the assistance of CINA-ASPECTS (Avicenna.AI, La Ciotat, France), a DLbased FDA-cleared and CE-marked algorithm designed to automatically compute ASPECTS. Differences were evaluated in both performance and interpretation time between the assisted and unassisted assessments.RESULTSWith software aid, readers demonstrated increased region-based accuracy from 72.4% to 76.5% (p<0.05), and increased ROC AUC from 0.749 to 0.788 (p<0.05). Notably, all readers exhibited an improved ROC AUC when utilizing the software. Moreover, use of the algorithm improved the score-based inter-observer reliability and correlation coefficient of ASPECTS evaluation by 0.222 and 0.087 (p<0.0001), respectively. Additionally, the readers' mean time spent analyzing a case was significantly reduced by 6% (p<0.05) when aided by the algorithm.CONCLUSIONSWith the assistance of the algorithm, readers' analyses were not only more accurate but also faster. Additionally, the overall ASPECTS evaluation exhibited greater consistency, less variabilities and higher precision compared to the reference standard. This novel tool has the potential to enhance patient selection for appropriate treatment by enabling physicians to deliver accurate and timely diagnosis of acute ischemic stroke.ABBREVIATIONSASPECTS = Alberta Stroke Program Early Computed Tomography Score; DL = Deep Learning; EIC = Early Ischemic Changes; ICC = Intraclass Correlation Coefficient; IS = Ischemic Stroke; ROC AUC = Receiver Operating Characteristics Area Under the Curve.
背景和目的ASPECTS 是急性缺血性卒中治疗的一个长期存在且有据可查的选择标准，然而，ASPECTS 的判读对医生来说是一项具有挑战性且耗时的任务，观察者之间存在显著差异。我们进行了一项多阅读器、多病例研究，让阅读器在无深度学习（DL）算法支持和有深度学习算法支持的情况下评估 ASPECTS，以分析该软件对临床医生的表现和解读时间的影响。参考标准由三位可获得基线 CTA 和 CTP 数据的神经放射学专家共同制定。随后，另外八名临床医生（四名典型的 ASPECTS 阅读器和四名资深神经放射科医生）分别在没有 CINA-ASPECTS (Avicenna.AI, La Ciotat, France) 的情况下和在 CINA-ASPECTS (Avicenna.AI, La Ciotat, France) 的协助下分析了 NCCT 扫描，CINA-ASPECTS 是一种基于 DL 的、经 FDA 批准和 CE 认证的算法，旨在自动计算 ASPECTS。结果在软件辅助下，读者基于区域的准确率从 72.4% 提高到 76.5%（P<0.05），ROC AUC 从 0.749 提高到 0.788（P<0.05）。值得注意的是，使用该软件后，所有读者的 ROC AUC 都有所提高。此外，使用该算法后，基于评分的观察者间可靠性和 ASPECTS 评估的相关系数分别提高了 0.222 和 0.087（p<0.0001）。结论在算法的帮助下，读者的分析不仅更加准确，而且速度更快。此外，与参考标准相比，ASPECTS 评估的整体一致性更高、变异性更小、精确度更高。这一新型工具可使医生准确、及时地诊断急性缺血性卒中，从而提高患者选择适当治疗的能力。 ABBREVIATIONSASPECTS = Alberta Stroke Program Early Computed Tomography Score；DL = Deep Learning；EIC = Early Ischemic Changes；ICC = Intraclass Correlation Coefficient；IS = Ischemic Stroke；ROC AUC = Receiver Operating Characteristics Area Under the Curve。
{"title":"Deep Learning-Based ASPECTS Algorithm Enhances Reader Performance and Reduces Interpretation Time.","authors":"Angela Ayobi,Adam Davis,Peter D Chang,Daniel S Chow,Kambiz Nael,Maxime Tassy,Sarah Quenet,Sylvain Fogola,Peter Shabe,David Fussell,Christophe Avare,Yasmina Chaibi","doi":"10.3174/ajnr.a8491","DOIUrl":"https://doi.org/10.3174/ajnr.a8491","url":null,"abstract":"BACKGROUND AND PURPOSEASPECTS is a long-standing and well documented selection criteria for acute ischemic stroke treatment, however, the interpretation of ASPECTS is a challenging and time-consuming task for physicians with significant interobserver variabilities. We conducted a multi-reader, multi-case study in which readers assessed ASPECTS without and with the support of a deep learning (DL)-based algorithm in order to analyze the impact of the software on clinicians' performance and interpretation time.MATERIALS AND METHODSA total of 200 NCCT scans from 5 clinical sites (27 scanner models, 4 different vendors) were retrospectively collected. Reference standard was established through the consensus of three expert neuroradiologists who had access to baseline CTA and CTP data. Subsequently, eight additional clinicians (four typical ASPECTS reader and four senior neuroradiologists) analyzed the NCCT scans without and with the assistance of CINA-ASPECTS (Avicenna.AI, La Ciotat, France), a DLbased FDA-cleared and CE-marked algorithm designed to automatically compute ASPECTS. Differences were evaluated in both performance and interpretation time between the assisted and unassisted assessments.RESULTSWith software aid, readers demonstrated increased region-based accuracy from 72.4% to 76.5% (p<0.05), and increased ROC AUC from 0.749 to 0.788 (p<0.05). Notably, all readers exhibited an improved ROC AUC when utilizing the software. Moreover, use of the algorithm improved the score-based inter-observer reliability and correlation coefficient of ASPECTS evaluation by 0.222 and 0.087 (p<0.0001), respectively. Additionally, the readers' mean time spent analyzing a case was significantly reduced by 6% (p<0.05) when aided by the algorithm.CONCLUSIONSWith the assistance of the algorithm, readers' analyses were not only more accurate but also faster. Additionally, the overall ASPECTS evaluation exhibited greater consistency, less variabilities and higher precision compared to the reference standard. This novel tool has the potential to enhance patient selection for appropriate treatment by enabling physicians to deliver accurate and timely diagnosis of acute ischemic stroke.ABBREVIATIONSASPECTS = Alberta Stroke Program Early Computed Tomography Score; DL = Deep Learning; EIC = Early Ischemic Changes; ICC = Intraclass Correlation Coefficient; IS = Ischemic Stroke; ROC AUC = Receiver Operating Characteristics Area Under the Curve.","PeriodicalId":7875,"journal":{"name":"American Journal of Neuroradiology","volume":null,"pages":null},"PeriodicalIF":3.5,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142196371","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}
Sang Hyo Lee,O-Ki Kwon,Young Deok Kim,Yongjae Lee,Chang Wan Oh,Jae Seung Bang,Si Un Lee,Min-Yong Kwon,Seung Pil Ban
BACKGROUND AND PURPOSESwelling of the salivary glands, known as contrast-induced sialadenitis (CIS), is an adverse reaction to iodide contrast agents. However, the incidence and risk factors of CIS after therapeutic neuroendovascular procedures have not yet been established.MATERIALS AND METHODSDemographic and procedural factors that may influence the development of CIS were retrospectively analyzed to identify the incidence and risk factors of this condition. A total of 780 patients who underwent therapeutic neuroendovascular procedures between January 1, 2022 and December 31, 2022 were investigated. The risk factors affecting CIS were analyzed using multivariate logistic regression, and the quantitative degree of association between the volume of contrast administered and occurrence of CIS was determined using the receiver operating characteristic (ROC) curve.RESULTSThe incidence of CIS after therapeutic neuroendovascular procedures was 4.2%. Multivariate logistic regression analysis showed that female sex (odds ratio [OR] = 4.420, 95% confidence interval [CI]: 1.377-14.190, p = 0.013), volume of contrast administered (OR = 1.007, 95% CI: 1.003-1.101, p < 0.001), and guiding catheter tip located within the external carotid artery (ECA) (OR = 8.701, 95% CI: 3.459-21.885, p < 0.001) were independently associated with CIS occurrence after therapeutic neuroendovascular procedures. The volume of contrast administered had an area under the ROC curve of 0.723 (95% CI:0.635-0.810; p < 0.001), and the optimal cut-off value of the volume of contrast administered was 205 cc (sensitivity: 0.49, specificity: 0.87).CONCLUSIONSWe observed CIS in 4.2% of our patients undergoing therapeutic neuroendovascular procedures. This represents a higher incidence than previously reported. Female sex, volume of contrast administered, and guiding catheter tip located within the ECA are associated with CIS incidence.ABBREVIATIONSAUC = area under the ROC curve; BMI = body mass index; CIS = contrast-induced sialadenitis; ECA = external cerebral artery; GFR = glomerular filtration rate; ROC = receiver operating characteristic.
{"title":"Incidence and risk factors of contrast-induced sialadenitis after therapeutic neuroendovascular procedures.","authors":"Sang Hyo Lee,O-Ki Kwon,Young Deok Kim,Yongjae Lee,Chang Wan Oh,Jae Seung Bang,Si Un Lee,Min-Yong Kwon,Seung Pil Ban","doi":"10.3174/ajnr.a8492","DOIUrl":"https://doi.org/10.3174/ajnr.a8492","url":null,"abstract":"BACKGROUND AND PURPOSESwelling of the salivary glands, known as contrast-induced sialadenitis (CIS), is an adverse reaction to iodide contrast agents. However, the incidence and risk factors of CIS after therapeutic neuroendovascular procedures have not yet been established.MATERIALS AND METHODSDemographic and procedural factors that may influence the development of CIS were retrospectively analyzed to identify the incidence and risk factors of this condition. A total of 780 patients who underwent therapeutic neuroendovascular procedures between January 1, 2022 and December 31, 2022 were investigated. The risk factors affecting CIS were analyzed using multivariate logistic regression, and the quantitative degree of association between the volume of contrast administered and occurrence of CIS was determined using the receiver operating characteristic (ROC) curve.RESULTSThe incidence of CIS after therapeutic neuroendovascular procedures was 4.2%. Multivariate logistic regression analysis showed that female sex (odds ratio [OR] = 4.420, 95% confidence interval [CI]: 1.377-14.190, p = 0.013), volume of contrast administered (OR = 1.007, 95% CI: 1.003-1.101, p < 0.001), and guiding catheter tip located within the external carotid artery (ECA) (OR = 8.701, 95% CI: 3.459-21.885, p < 0.001) were independently associated with CIS occurrence after therapeutic neuroendovascular procedures. The volume of contrast administered had an area under the ROC curve of 0.723 (95% CI:0.635-0.810; p < 0.001), and the optimal cut-off value of the volume of contrast administered was 205 cc (sensitivity: 0.49, specificity: 0.87).CONCLUSIONSWe observed CIS in 4.2% of our patients undergoing therapeutic neuroendovascular procedures. This represents a higher incidence than previously reported. Female sex, volume of contrast administered, and guiding catheter tip located within the ECA are associated with CIS incidence.ABBREVIATIONSAUC = area under the ROC curve; BMI = body mass index; CIS = contrast-induced sialadenitis; ECA = external cerebral artery; GFR = glomerular filtration rate; ROC = receiver operating characteristic.","PeriodicalId":7875,"journal":{"name":"American Journal of Neuroradiology","volume":null,"pages":null},"PeriodicalIF":3.5,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142196345","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}
Chunwei Ying,Yasheng Chen,Yan Yan,Shaney Flores,Richard Laforest,Tammie L S Benzinger,Hongyu An
BACKGROUND AND PURPOSEIntegrated PET/MR allows the simultaneous acquisition of PET biomarkers and structural and functional MRI to study Alzheimer disease (AD). Attenuation correction (AC), crucial for PET quantification, can be performed using a deep learning approach, DL-Dixon, based on standard Dixon images. Longitudinal amyloid PET imaging, which provides important information about disease progression or treatment responses in AD, is usually acquired over several years. Hardware and software upgrades often occur during a multiple-year study period, resulting in data variability. This study aims to harmonize PET/MR DL-Dixon AC amid software and head coil updates and evaluate its accuracy and longitudinal consistency.MATERIALS AND METHODSTri-modality PET/MR and CT images were obtained from 329 participants, with a subset of 38 undergoing tri-modality scans twice within approximately three years. Transfer learning was employed to fine-tune DL-Dixon models on images from two scanner software versions (VB20P and VE11P) and two head coils (16-channel and 32-channel coils). The accuracy and longitudinal consistency of the DL-Dixon AC were evaluated. Power analyses were performed to estimate the sample size needed to detect various levels of longitudinal changes in the PET standardized uptake value ratio (SUVR).RESULTSThe DL-Dixon method demonstrated high accuracy across all data, irrespective of scanner software versions and head coils. More than 95.6% of brain voxels showed less than 10% PET relative absolute error in all participants. The median [interquartile range] PET mean relative absolute error was 1.10% [0.93%, 1.26%], 1.24% [1.03%, 1.54%], 0.99% [0.86%, 1.13%] in the cortical summary region, and 1.04% [0.83%, 1.36%], 1.08% [0.84%, 1.34%], 1.05% [0.72%, 1.32%] in cerebellum using the DL-Dixon models for the VB20P-16-channel-coil, VE11P-16-channel-coil and VE11P-32-channel-coil data, respectively. The within-subject coefficient of variation and intra-class correlation coefficient of PET SUVR in the cortical regions were comparable between the DL-Dixon and CT AC. Power analysis indicated that similar numbers of participants would be needed to detect the same level of PET changes using DL-Dixon and CT AC.CONCLUSIONSDL-Dixon exhibited excellent accuracy and longitudinal consistency across the two software versions and head coils, demonstrating its robustness for longitudinal PET/MR neuroimaging studies in AD.ABBREVIATIONSAC = attenuation correction; AD = Alzheimer disease; HU = Hounsfield unit; ICC = intraclass correlation coefficient; MAE = mean absolute error; MRAE = mean relative absolute error; pCT = pseudo-CT; PiB = Pittsburgh Compound B; SD = standard deviation; SUVR = standardized uptake value ratio; wCV = within-subject coefficient of variation.
背景和目的正电子发射计算机断层显像（PET）/磁共振成像（MRI）整合技术可同时获取正电子发射计算机断层显像生物标记物以及结构和功能磁共振成像，用于研究阿尔茨海默病（AD）。衰减校正（AC）对正电子发射计算机断层显像（PET）的量化至关重要，可使用基于标准 Dixon 图像的深度学习方法 DL-Dixon 进行。纵向淀粉样蛋白 PET 成像可提供有关 AD 疾病进展或治疗反应的重要信息，通常需要数年时间才能获得。在长达数年的研究期间，硬件和软件经常会升级，从而导致数据的变化。本研究旨在统一 PET/MR DL-Dixon AC amid 软件和头线圈更新，并评估其准确性和纵向一致性。材料和方法从 329 名参与者中获得了三模态 PET/MR 和 CT 图像，其中 38 人在大约三年内接受了两次三模态扫描。采用迁移学习对两种扫描仪软件版本（VB20P 和 VE11P）和两种头部线圈（16 通道和 32 通道线圈）的图像进行 DL-Dixon 模型微调。对 DL-Dixon AC 的准确性和纵向一致性进行了评估。结果无论扫描仪软件版本和头部线圈如何，DL-Dixon 方法在所有数据中都表现出很高的准确性。在所有参与者中，95.6%以上的大脑体素显示 PET 相对绝对误差小于 10%。PET 平均相对绝对误差的中位数[四分位间范围]为 1.10% [0.93%, 1.26%]，皮质汇总区为 1.24% [1.03%, 1.54%]，0.99% [0.86%, 1.13%]，1.04% [0.83%, 1.对 VB20P-16-channel-coil, VE11P-16-channel-coil 和 VE11P-32-channel-coil 数据使用 DL-Dixon 模型，小脑分别为 1.04% [0.83%, 1.36%]、1.08% [0.84%, 1.34%]、1.05% [0.72%, 1.32%]。皮质区域 PET SUVR 的受试者内变异系数和类内相关系数在 DL-Dixon 和 CT AC 之间具有可比性。结论DL-Dixon在两个软件版本和头线圈上都表现出了极好的准确性和纵向一致性，证明了它在AD的纵向PET/MR神经影像研究中的稳健性。缩略语AC=衰减校正；AD=阿尔茨海默病；HU=Hounsfield单位；ICC=类内相关系数；MAE=平均绝对误差；MRAE=平均相对绝对误差；pCT=伪CT；PiB=匹兹堡化合物B；SD=标准差；SUVR=标准化摄取值比；wCV=受试者内变异系数。
{"title":"Accuracy and longitudinal consistency of PET/MR attenuation correction in amyloid PET imaging amid software and hardware upgrades.","authors":"Chunwei Ying,Yasheng Chen,Yan Yan,Shaney Flores,Richard Laforest,Tammie L S Benzinger,Hongyu An","doi":"10.3174/ajnr.a8490","DOIUrl":"https://doi.org/10.3174/ajnr.a8490","url":null,"abstract":"BACKGROUND AND PURPOSEIntegrated PET/MR allows the simultaneous acquisition of PET biomarkers and structural and functional MRI to study Alzheimer disease (AD). Attenuation correction (AC), crucial for PET quantification, can be performed using a deep learning approach, DL-Dixon, based on standard Dixon images. Longitudinal amyloid PET imaging, which provides important information about disease progression or treatment responses in AD, is usually acquired over several years. Hardware and software upgrades often occur during a multiple-year study period, resulting in data variability. This study aims to harmonize PET/MR DL-Dixon AC amid software and head coil updates and evaluate its accuracy and longitudinal consistency.MATERIALS AND METHODSTri-modality PET/MR and CT images were obtained from 329 participants, with a subset of 38 undergoing tri-modality scans twice within approximately three years. Transfer learning was employed to fine-tune DL-Dixon models on images from two scanner software versions (VB20P and VE11P) and two head coils (16-channel and 32-channel coils). The accuracy and longitudinal consistency of the DL-Dixon AC were evaluated. Power analyses were performed to estimate the sample size needed to detect various levels of longitudinal changes in the PET standardized uptake value ratio (SUVR).RESULTSThe DL-Dixon method demonstrated high accuracy across all data, irrespective of scanner software versions and head coils. More than 95.6% of brain voxels showed less than 10% PET relative absolute error in all participants. The median [interquartile range] PET mean relative absolute error was 1.10% [0.93%, 1.26%], 1.24% [1.03%, 1.54%], 0.99% [0.86%, 1.13%] in the cortical summary region, and 1.04% [0.83%, 1.36%], 1.08% [0.84%, 1.34%], 1.05% [0.72%, 1.32%] in cerebellum using the DL-Dixon models for the VB20P-16-channel-coil, VE11P-16-channel-coil and VE11P-32-channel-coil data, respectively. The within-subject coefficient of variation and intra-class correlation coefficient of PET SUVR in the cortical regions were comparable between the DL-Dixon and CT AC. Power analysis indicated that similar numbers of participants would be needed to detect the same level of PET changes using DL-Dixon and CT AC.CONCLUSIONSDL-Dixon exhibited excellent accuracy and longitudinal consistency across the two software versions and head coils, demonstrating its robustness for longitudinal PET/MR neuroimaging studies in AD.ABBREVIATIONSAC = attenuation correction; AD = Alzheimer disease; HU = Hounsfield unit; ICC = intraclass correlation coefficient; MAE = mean absolute error; MRAE = mean relative absolute error; pCT = pseudo-CT; PiB = Pittsburgh Compound B; SD = standard deviation; SUVR = standardized uptake value ratio; wCV = within-subject coefficient of variation.","PeriodicalId":7875,"journal":{"name":"American Journal of Neuroradiology","volume":null,"pages":null},"PeriodicalIF":3.5,"publicationDate":"2024-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142196372","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：