Radiology advances最新文献

{"title":"Quantifying lower extremity blood flow using low-dose CT perfusion: validation in a swine model.","authors":"Alireza Shojazadeh, Negin Hadjiabdolhamid, Dale J Black, Ines Antunes, Chaeeun Lee, Wenbo Li, Sabee Molloi","doi":"10.1093/radadv/umae029","DOIUrl":"10.1093/radadv/umae029","url":null,"abstract":"<p><strong>Background: </strong>Quantitative assessment of blood flow in peripheral extremities in conjunction with simultaneous CT angiography measurements can improve risk assessment and provide a critical decision-making tool for patients across a wide spectrum of vascular disease severity.</p><p><strong>Purpose: </strong>This study assessed the reproducibility and accuracy of lower extremity blood flow measurements with a low-dose first-pass analysis CT perfusion technique.</p><p><strong>Materials and methods: </strong>This prospective study utilized 16 Yorkshire Swine to obtain lower extremity blood flow CT measurements at baseline and under induced femoral stenosis using a vascular occluder. Thirty-three pairs of CT measurements evaluated reproducibility, and 43 CT measurements assessed accuracy against ultrasound flow probe references. Contrast agent and saline chaser were both injected peripherally at a rate of 5 mL/s. Bolus tracking was used, and a pre-contrast and post-contrast helical scan were acquired at the base and approximately the peak of the femoral enhancement (CT angiogram), respectively. The acquired data were then used as analytical inputs into a first-pass analysis model to derive perfusion in mL/min/g. The reproducibility and accuracy of lower extremity perfusion measurements were assessed via Mixed model regression and Bland-Altman analysis.</p><p><strong>Results: </strong>Calculated CT perfusion measurements derived from first-pass analysis technique (<i>P</i> _CT), and the reference standard ultrasound perfusion measurements (<i>P</i> _ref) were related by <i>P</i> _CT = 1.06 <i>P</i> _ref + 0.00 (<i>r</i> ² = 0.90, Root-Mean-Square Error [RMSE] = 0.01 mL/min/g). The first (<i>P</i> ₁) and second (<i>P</i> ₂) CT perfusion measurements were related by <i>P</i> ₂ = 0.98 <i>P</i> ₁ + 0.02 (<i>r</i> = 0.97, RMSE = 0.11 mL/min/g). The average effective dose of perfusion measurement using first-pass analysis technique was calculated to be only 2.13 mSv.</p><p><strong>Conclusion: </strong>The low-dose quantitative CT perfusion technique can accurately measure lower extremity perfusion (mL/min/g) using only 2 helical scans. The CT angiogram and perfusion measurements can be used as a comprehensive technique for morphological and physiological assessment of limb ischemia.</p>","PeriodicalId":519940,"journal":{"name":"Radiology advances","volume":"1 4","pages":"umae029"},"PeriodicalIF":0.0,"publicationDate":"2024-11-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12429254/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145246334","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":"Deep learning assessment of disproportionately enlarged subarachnoid-space hydrocephalus in Hakim's disease or idiopathic normal pressure hydrocephalus.","authors":"Shigeki Yamada, Hirotaka Ito, Chifumi Iseki, Toshiyuki Kondo, Tomoyasu Yamanaka, Motoki Tanikawa, Tomohiro Otani, Satoshi Ii, Yasuyuki Ohta, Yoshiyuki Watanabe, Shigeo Wada, Marie Oshima, Mitsuhito Mase","doi":"10.1093/radadv/umae027","DOIUrl":"10.1093/radadv/umae027","url":null,"abstract":"<p><strong>Background: </strong>Disproportionately enlarged subarachnoid-space hydrocephalus (DESH) is a key feature of Hakim's disease (synonymous with idiopathic normal pressure hydrocephalus; iNPH). However, it previously had been only subjectively evaluated.</p><p><strong>Purpose: </strong>This study aims to evaluate the usefulness of MRI indices, derived from deep learning segmentation of cerebrospinal fluid (CSF) spaces, for DESH detection and to establish their optimal thresholds.</p><p><strong>Materials and methods: </strong>This study retrospectively enrolled a total of 1009 participants, including 77 patients diagnosed with Hakim's disease, 380 healthy volunteers, 163 with mild cognitive impairment, 256 with Alzheimer's disease, and 217 with other types of neurodegenerative diseases. DESH, ventriculomegaly, tightened sulci in the high convexities, and Sylvian fissure dilatation were evaluated on three-dimensional T1-weighted MRI by radiologists. The total ventricles, high-convexity part of the subarachnoid space, and Sylvian fissure and basal cistern were automatically segmented using the CSF Space Analysis application (FUJIFILM Corporation). Moreover, DESH, Venthi, and Sylhi indices were calculated based on these 3 regions. The area under the receiver-operating characteristic curves of these indices and region volumes (volume ratios) for DESH detection were calculated.</p><p><strong>Results: </strong>Of the 1009 participants, 101 (10%) presented with DESH. The DESH, Venthi, and Sylhi indices performed well with 95.0%-96.0% sensitivity and 91.5%-96.8% specificity at optimal thresholds. All patients with Hakim's disease were diagnosed with DESH, despite variations in severity. In patients with Hakim's disease, with or without Alzheimer's disease, the DESH index and total ventricular volume were significantly higher compared to patients with Alzheimer's disease, although the total intracranial cerebrospinal fluid volume was significantly lower.</p><p><strong>Conclusion: </strong>DESH, Venthi, and Sylhi indices, and the volumes and volume ratios of the ventricle and high-convexity part of the subarachnoid space computed using deep learning were useful for the DESH detection that may help to improve the diagnosis of Hakim's disease (ie, iNPH).</p>","PeriodicalId":519940,"journal":{"name":"Radiology advances","volume":"1 3","pages":"umae027"},"PeriodicalIF":0.0,"publicationDate":"2024-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12429205/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145246329","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":"Quantification of myocardial oxygen extraction fraction on noncontrast MRI enabled by deep learning.","authors":"Ran Li, Cihat Eldeniz, Keyan Wang, Natalie Nguyen, Thomas H Schindler, Qi Huang, Linda R Peterson, Yang Yang, Yan Yan, Jingliang Cheng, Pamela K Woodard, Jie Zheng","doi":"10.1093/radadv/umae026","DOIUrl":"10.1093/radadv/umae026","url":null,"abstract":"<p><strong>Purpose: </strong>To develop a new deep learning enabled cardiovascular magnetic resonance (CMR) approach for noncontrast quantification of myocardial oxygen extraction fraction (mOEF) and myocardial blood volume (MBV) in vivo.</p><p><strong>Materials and methods: </strong>An asymmetric spin-echo prepared CMR sequence was created in a 3 T MRI clinical system. A UNet-based fully connected neural network was developed based on a theoretical model of CMR signals to calculate mOEF and MBV. Twenty healthy volunteers (20-30 years old, 11 females) underwent CMR scans at 3 short-axial slices (16 myocardial segments) on 2 different days. The reproducibility was assessed by the coefficient of variation. Ten patients with chronic myocardial infarction were examined to evaluate the feasibility of this CMR method to detect abnormality of mOEF and MBV.</p><p><strong>Results: </strong>Among the volunteers, the average global mOEF and MBV on both days was 0.58 ± 0.07 and 9.5% ± 1.5%, respectively, which agreed well with data measured by other imaging modalities. The coefficient of variation of mOEF was 8.4%, 4.5%, and 2.6%, on a basis of segment, slice, and participant, respectively. No significant difference in mOEF was shown among 3 slices or among different myocardial segments. Female participants showed significantly higher segmental mOEF than male participants (<i>P</i> < .001). Regional mOEF decrease 40% in CMR-confirmed myocardial infarction core, compared to normal myocardial regions.</p><p><strong>Conclusion: </strong>The new deep learning-enabled CMR approach allows noncontrast quantification of mOEF and MBV with good to excellent reproducibility. This technique could provide an objective contrast-free means to assess and serially measure hypoxia-relief effects of therapeutic interventional strategies to save viable myocardial tissues.</p>","PeriodicalId":519940,"journal":{"name":"Radiology advances","volume":"1 4","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12245170/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144610887","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":"Magnetic particle imaging enables nonradioactive quantitative sentinel lymph node identification: feasibility proof in murine models.","authors":"Olivia C Sehl, Kelvin Guo, Abdul Rahman Mohtasebzadeh, Petrina Kim, Benjamin Fellows, Marcela Weyhmiller, Patrick W Goodwill, Max Wintermark, Stephen Y Lai, Paula J Foster, Joan M Greve","doi":"10.1093/radadv/umae024","DOIUrl":"10.1093/radadv/umae024","url":null,"abstract":"<p><strong>Background: </strong>Sentinel lymph node biopsy (SLNB) is an important cancer diagnostic staging procedure. Conventional SLNB procedures with ^99mTc radiotracers and scintigraphy are constrained by tracer half-life and, in some cases, insufficient image resolution. Here, we explore an alternative magnetic (nonradioactive) image-guided SLNB procedure.</p><p><strong>Purpose: </strong>To demonstrate that magnetic particle imaging (MPI) lymphography can sensitively, specifically, and quantitatively identify and map sentinel lymph modes (SLNs) in murine models in multiple regional lymphatic basins.</p><p><strong>Materials and methods: </strong>Iron oxide nanoparticles were administered intradermally to healthy C57BL/6 mice (male, 12-week-old, n = 5). The nanoparticles (0.675 mg Fe/kg) were injected into the tongue, forepaw, base of tail, or hind footpad, then detected by 3-dimensional MPI at multiple timepoints between 1 hour and 4 to 6 days. In this mouse model, the SLN is represented by the first lymph node draining from the injection site. SLNs were extracted to verify the MPI signal ex vivo and processed using Perl's Prussian iron staining. Paired <i>t</i>-test was conducted to compare MPI signal from SLNs in vivo vs. ex vivo and considered significant if <i>P</i> < .05.</p><p><strong>Results: </strong>MPI lymphography identified SLNs in multiple lymphatic pathways, including the cervical SLN draining the tongue, axillary SLN draining the forepaw, inguinal SLN draining the tail, and popliteal SLN draining the footpad. MPI signal in lymph nodes was present after 1 hour and stable for the duration of the study (4-6 days). Perl's Prussian iron staining was identified in the subcapsular space of excised SLNs.</p><p><strong>Conclusion: </strong>Our data support the use of MPI lymphography to specifically detect SLN(s) using a magnetic tracer for a minimum of 4 to 6 days, thereby providing information required to plan the SLN approach in cancer surgery. As clinical-scale MPI is developed, translation will benefit from a history of using iron-oxide nanoparticles in human imaging and recent regulatory-approvals for use in SLNB.</p>","PeriodicalId":519940,"journal":{"name":"Radiology advances","volume":"1 3","pages":"umae024"},"PeriodicalIF":0.0,"publicationDate":"2024-10-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11576474/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142690348","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":"Pulmonary embolism detection without intravenous contrast using electron density and Z-effective maps from dual-energy CT.","authors":"Tommaso D'Angelo, Simone Barbera, Velio Ascenti, Giuseppe Cicero, Simone Terrani, Damiano Caruso, Andrea Laghi, Federico Fontana, Massimo Venturini, Filippo Piacentino, Christian Booz, Thomas J Vogl, Ibrahim Yel, Maria Adele Marino, Silvio Mazziotti, Giorgio Ascenti","doi":"10.1093/radadv/umae025","DOIUrl":"10.1093/radadv/umae025","url":null,"abstract":"<p><strong>Purpose: </strong>This study aims to evaluate the feasibility of using electron density (ED) maps combined with Z-effective (Zeff) images obtained from unenhanced dual-layer dual-energy CT (dl-DECT) scans of the chest for the detection of pulmonary embolism (PE).</p><p><strong>Materials and methods: </strong>A retrospective analysis was conducted on consecutive patients who underwent for contrast-enhanced chest CT (CECT) clinically suspected of PE or acute aortic syndrome. These scans were performed on a single dl-DECT scanner between October 2021 and November 2023. To distinguish emboli from circulating blood, color-coded maps were generated from the ED dataset superimposed on Zeff images, which were acquired from the unenhanced phase. Two radiologists with different levels of expertise independently assessed the presence of PE in the generated ED-Zeff maps, blinded to CECT results, which served as the reference standard. Diagnostic accuracy of ED-Zeff maps was assessed for each reader.</p><p><strong>Results: </strong>The final study cohort comprised 150 patients, with 92 males (mean age: 68 ± 10 years, range: 47-93 years) and 58 females (mean age: 66 ± 15 years, range 38-89 years). ED-Zeff maps demonstrated high diagnostic performance, yielding accuracy, sensitivity, and specificity, respectively, of 86.67% (113/150, 95% CI, 80.16%-91.66%), 85% (17/20, 95% CI, 79.89%-92.19%), and 86.92% (113/130, 95% CI, 79.89%-92.19%). Ed-Zeff maps were able to identify PE in 85% of positive cases. Cohen's kappa coefficient indicated excellent intra- and interobserver agreement (κ ≥ 0.9).</p><p><strong>Conclusion: </strong>ED maps combined with Zeff images from unenhanced dl-DECT scans represent a feasible tool for detecting PE and may prove useful in evaluating patients with contraindications to iodinated contrast.</p>","PeriodicalId":519940,"journal":{"name":"Radiology advances","volume":"1 3","pages":"umae025"},"PeriodicalIF":0.0,"publicationDate":"2024-10-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12429230/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145246326","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":"Different sodium concentrations of noncancerous and cancerous prostate tissue seen on MRI using an external coil.","authors":"Josephine L Tan, Vibhuti Kalia, Stephen E Pautler, Glenn Bauman, Lena V Gast, Max Müller, Armin M Nagel, Jonathan D Thiessen, Timothy J Scholl, Alireza Akbari","doi":"10.1093/radadv/umae023","DOIUrl":"10.1093/radadv/umae023","url":null,"abstract":"<p><strong>Background: </strong>Sodium (²³Na) MRI of prostate cancer (PCa) is a novel but underdocumented technique conventionally acquired using an endorectal coil. These endorectal coils are associated with challenges (e.g., a nonuniform sensitivity profile, limited prostate coverage, patient discomfort) that could be mitigated with an external ²³Na MRI coil.</p><p><strong>Purpose: </strong>To quantify tissue sodium concentration (TSC) differences within the prostate of participants with PCa and healthy volunteers using an external ²³Na MRI radiofrequency coil at 3 T.</p><p><strong>Materials and methods: </strong>A prospective study was conducted from January 2022 to June 2024 in healthy volunteers and participants with biopsy-proven PCa. Prostate ²³Na MRI was acquired on a 3-T PET/MRI scanner using a custom-built 2-loop (diameter, 18 cm) butterfly surface coil tuned for the ²³Na frequency (32.6 MHz). The percent difference in TSC (ΔTSC) between prostate cancer lesions and surrounding noncancerous prostate tissue of the peripheral zone (PZ) and transition zone (TZ) was evaluated using a 1-sample <i>t</i>-test. TSC was compared to apparent diffusion coefficient (ADC) measurements as a clinical reference.</p><p><strong>Results: </strong>Six healthy volunteers (mean age, 54.5 years ± 12.7) and 20 participants with PCa (mean age, 70.7 years ± 8.3) were evaluated. A total of 31 lesions were detected (21 PZ, 10 TZ) across PCa participants. Compared to noncancerous prostate tissue, prostate cancer lesions had significantly lower TSC (ΔTSC, -14.1% ± 18.2, <i>P</i> = .0002) and ADC (ΔADC, -26.6% ± 18.7, <i>P</i> < .0001).</p><p><strong>Conclusion: </strong>We used an external ²³Na MRI coil for whole-gland comparison of TSC in PCa and noncancerous prostate tissue at 3 T. PCa lesions presented with lower TSC compared to surrounding noncancerous PZ and TZ tissue. These findings demonstrate the feasibility of an external ²³Na MRI coil to quantify TSC in the prostate and offer a promising, noninvasive approach to PCa diagnosis and management.</p>","PeriodicalId":519940,"journal":{"name":"Radiology advances","volume":"1 3","pages":"umae023"},"PeriodicalIF":0.0,"publicationDate":"2024-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11578593/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142690346","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":"Reducing motion artifacts in craniocervical background subtraction angiography with deformable registration and unsupervised deep learning.","authors":"Chaochao Zhou, Ramez N Abdalla, Dayeong An, Syed H A Faruqui, Teymour Sadrieh, Mohayad Alzein, Rayan Nehme, Ali Shaibani, Sameer A Ansari, Donald R Cantrell","doi":"10.1093/radadv/umae020","DOIUrl":"10.1093/radadv/umae020","url":null,"abstract":"<p><strong>Background: </strong>In clinical practice, digital subtraction angiography (DSA) often suffers from misregistration artifact resulting from voluntary, respiratory, and cardiac motion during acquisition. Most prior efforts to register the background DSA mask to subsequent postcontrast images rely on key point registration using iterative optimization, which has limited real-time application.</p><p><strong>Purpose: </strong>Leveraging state-of-the-art, unsupervised deep learning, we aim to develop a fast, deformable registration model to substantially reduce DSA misregistration in craniocervical angiography without compromising spatial resolution or introducing new artifacts.</p><p><strong>Materials and methods: </strong>We extend HyperMorph, an open source deep learning deformable registration framework, to reduce motion artifacts in DSA. Novel image similarity loss functions with vessel layer estimation were introduced to optimize background registration, making it robust to the variable presence of intravascular iodinated contrast.</p><p><strong>Results: </strong>A total of 516 studies with 5,240 angiographic series were collected and divided into training (5,046 series) and hold-out test (194 series) sets. Blinded algorithm rankings and Likert scores on 5-point scales (1 = worst, 5 = best) were generated by 3 practicing interventional neuroradiologists using 50 series randomly selected from the hold-out test set. Compared to traditional DSA, our learning-based background subtraction angiography (BSA) significant improved vascular fidelity (2.4 ± 0.6 for DSA vs. 3.6 ± 0.5 for BSA), subtraction artifacts (2.0 ± 0.4 for DSA vs. 3.9 ± 0.3 for BSA), and overall quality (2.1 ± 0.5 for DSA vs. 3.9 ± 0.4 for BSA) (<i>P</i> < .0001). Learning-based BSA also significantly outperformed affine registration-based BSA (<i>P</i> < .0001). The average inference time for learning-based BSA was 30 milliseconds per frame on our hardware.</p><p><strong>Conclusion: </strong>The results demonstrate that deep learning deformable registration, combined with an appropriate loss function, can significantly reduce the motion artifacts that degrade DSA.</p>","PeriodicalId":519940,"journal":{"name":"Radiology advances","volume":"1 3","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12416918/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145031620","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 image-domain deep-learning denoising technique for accelerated parallel brain MRI: prospective clinical evaluation.","authors":"Laura Onac, Lorand Dobai, Andrei Mouraviev, Mihai A Badila, Denise Yap, Samantha Kennedy, Annie Nguyen, Emi Gal, Daniel K Sodickson","doi":"10.1093/radadv/umae022","DOIUrl":"10.1093/radadv/umae022","url":null,"abstract":"<p><strong>Background: </strong>Parallel imaging can accelerate MRI acquisitions, but excessive accelerations can introduce amplified noise and aliasing artifacts.</p><p><strong>Purpose: </strong>To evaluate a vendor-agnostic AI-based approach to remove image degradation artifacts in highly accelerated MRI scans, improving image quality and reducing scan time.</p><p><strong>Materials and methods: </strong>Training was performed by retrospectively degrading standard accelerated images. Evaluation was performed with both retrospective and prospectively-collected highly accelerated images. Retrospective data were taken from ∼2000 MRI studies obtained between August 2016 and October 2022, and prospective data were collected in >200 subjects between June and November 2022, using scanners from multiple vendors and locations. Scan time data were collected from prospective studies and used to compute time savings per sequence and per protocol for each vendor. Images were evaluated qualitatively by 5 board-certified radiologists and quantitatively by assessing noise, contrast, and spatial resolution. Paired Wilcoxon signed-rank tests were used to compare model outputs to model inputs and low-acceleration images.</p><p><strong>Results: </strong>Images from 101 adults from 5 sites and 6 scanner models from different vendors were enrolled. 89% of imaged subjects had noteworthy imaging features or pathology. Model outputs were rated superior to model inputs (<i>P < .</i>001) and most were either non-inferior (<i>P</i> _inf <i> > </i>.05) or superior (<i>P</i> _sup <i> < </i>.05) to baseline images in qualitative metrics of image quality and feature visibility. Quantitative evaluation of signal-to-noise ratio and contrast-to-noise ratio improved for model outputs compared to inputs (<i>P < .</i>001) or baseline images (<i>P < .</i>005). Apparent resolution measured using the full width at half maximum or minimum was either enhanced (<i>P</i> _sup <i> < </i>.05) or preserved (non-superior <i>P</i> _sup <i> > </i>.05 and non-inferior <i>P</i> _inf <i> > </i>.05). The scan time was reduced by an average of 29% (19%-41% per sequence).</p><p><strong>Conclusion: </strong>This vendor-agnostic AI-based method achieved robust scan time savings without loss of image quality, potentially allowing for reduced cost and improved patient experience.</p>","PeriodicalId":519940,"journal":{"name":"Radiology advances","volume":"1 3","pages":"umae022"},"PeriodicalIF":0.0,"publicationDate":"2024-08-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12481681/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145246282","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":"¹⁸F-FDG-PET-based deep learning for predicting cognitive decline in non-demented elderly across the Alzheimer's disease clinical spectrum.","authors":"Beomseok Sohn, Seok Jong Chung, Jeong Ryong Lee, Dosik Hwang, Wanying Xie, Ling Ling Chan, Yoon Seong Choi","doi":"10.1093/radadv/umae021","DOIUrl":"10.1093/radadv/umae021","url":null,"abstract":"<p><strong>Background: </strong>With disease-modifying treatments for Alzheimer's disease (AD), prognostic tools for the pre-dementia stage are needed. This study aimed to evaluate the prognostic value of an ¹⁸F-fluorodeoxyglucose-positron emission tomography (¹⁸F-FDG-PET)-based deep-learning (DL) model in the pre-dementia stage of mild cognitive impairment (MCI) and normal cognition (NC).</p><p><strong>Materials and methods: </strong>A ¹⁸F-FDG-PET-based DL model was developed to classify diagnosis of AD-dementia vs NC using AD Neuroimaging Initiative (ADNI) and Japanese-ADNI (J-ADNI) datasets (<i>n</i> = 756), which provided the degree of similarity to AD-dementia. The prognostic value of the DL output for cognitive decline was assessed in the ADNI MCI (<i>n</i> = 663), J-ADNI MCI (<i>n</i> = 129), and Harvard Aging Brain Study (HABS) NC (<i>n</i> = 274) participants using Cox regression and calculating the integrated area under the time-dependent ROC curves (iAUC), along with clinical information and ¹⁸F-FDG-PET standardized uptake value ratio (SUVR). Subgroup analysis in the amyloid-positive ADNI MCI participants was performed using Cox regression and calculating the area under the time-dependent ROC (tdAUC) curves at 4-year follow-up to assess prognostic value of DL output over clinical information, ¹⁸F-FDG-PET SUVR, and amyloid PET Centiloids.</p><p><strong>Results: </strong>DL output remained independently prognostic among other factors in all three datasets (<i>P</i> < .05 for all by Cox regression). By adding DL output to other prognostic factors, prediction significantly improved in ADNI-MCI (iAUC differences 0.020 [0.007-0.034] before and after adding DL output) and improved without statistical significance in J-ADNI (0.020 [-0.005 to 0.044], and HABS-NC sets (0.059 [-0.003 to 0.126]). DL output showed independent (<i>P</i> = .002 by Cox regression) and significant added prognostic value (tdROC difference 0.019 [<0.001-0.036]) over clinical information, ¹⁸F-FDG-PET SUVR, and Centiloids in the amyloid-positive ADNI MCI participants.</p><p><strong>Conclusion: </strong>The ¹⁸F-FDG-PET-based DL model demonstrated the potential to improve cognitive decline prediction beyond clinical information, and conventional measures from ¹⁸F-FDG-PET and amyloid PET and may prove useful for clinical trial recruitment and individualized management.</p>","PeriodicalId":519940,"journal":{"name":"Radiology advances","volume":"1 3","pages":"umae021"},"PeriodicalIF":0.0,"publicationDate":"2024-08-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12429239/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145246279","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":"Deep learning-based amyloid PET harmonization to predict cognitive decline in non-demented elderly.","authors":"Yoon Seong Choi, Pei Ing Ngam, Jeong Ryong Lee, Dosik Hwang, Eng-King Tan","doi":"10.1093/radadv/umae019","DOIUrl":"10.1093/radadv/umae019","url":null,"abstract":"<p><strong>Background: </strong>The robustness of conventional amyloid PET harmonization across tracers has been questioned.</p><p><strong>Purpose: </strong>To evaluate deep learning-based harmonization of amyloid PET in predicting conversion from cognitively unimpaired (CU) to mild cognitive impairment (MCI) and MCI to Alzheimer's disease (AD).</p><p><strong>Materials and methods: </strong>We developed an amyloid PET-based deep-learning model to classify participants with a clinical diagnosis of AD-dementia vs CU across different tracers from the Alzheimer's Disease Neuroimaging Initiative (ADNI), Japanese ADNI, and Australian Imaging, Biomarker, and Lifestyle cohorts (<i>n</i> = 1050). The model output [deep learning-based probability of Alzheimer's disease-dementia (DL-ADprob)], with other prognostic factors, was evaluated for predicting cognitive decline in ADNI-MCI (<i>n</i> = 451) and Harvard Aging Brain Study (HABS)-CU (<i>n</i> = 271) participants using Cox regression and area under time-dependent receiver operating characteristics curve (tdAUC) at 4-year follow-up. Subgroup analyses were performed in the ADNI-MCI group for conversion from amyloid-positive to AD and from amyloid negative to positive. Intraclass correlation coefficient (ICC) of DL-ADprob between tracers was calculated in the Global Alzheimer's Association Interactive Network dataset (<i>n</i> = 155).</p><p><strong>Results: </strong>DL-ADprob was independently prognostic in both ADNI-MCI (<i>P</i> < .001) and HABS-CU (<i>P</i> = .048) sets. Adding DL-ADprob to other factors increased prognostic performances in both ADNI-MCI (tdAUC 0.758 [0.721-0.792] vs 0.782 [0.742-0.818], tdAUC difference 0.023 [0.007-0.038]) and HABS-CU (tdAUC 0.846 [0.755-0.925] vs 0.870 [0.773-0.943], tdAUC difference 0.022 [-0.004 to 0.053]). DL-ADprob was independently prognostic in amyloid-positive (<i>P</i> < .001) and amyloid-negative subgroups (<i>P</i> = .007). DL-ADprob showed incremental prognostic value in amyloid-positive (tdAUC 0.666 [0.623-0.713] vs 0.706 [0.657-0.755], tdAUC difference 0.039 [0.016-0.064]), but not in amyloid-negative (tdAUC 0.818 [0.757-0.882] vs 0.816 [0.751-0.880], tdAUC difference -0.002 [-0.031 to 0.029]) subgroup. The pairwise ICCs of DL-ADprob between Pittsburgh compound B and florbetapir, florbetaben, and flutemetamol, respectively, ranged from 0.913 to 0.935.</p><p><strong>Conclusion: </strong>Deep learning-based harmonization of amyloid PET improves cognitive decline prediction in non-demented elderly, suggesting it could complement conventional amyloid PET measures.</p>","PeriodicalId":519940,"journal":{"name":"Radiology advances","volume":"1 2","pages":"umae019"},"PeriodicalIF":0.0,"publicationDate":"2024-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12429267/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145246146","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}