Data Extraction from Free-Text Reports on Mechanical Thrombectomy in Acute Ischemic Stroke Using ChatGPT: A Retrospective Analysis.

Background Procedural details of mechanical thrombectomy in patients with ischemic stroke are important predictors of clinical outcome and are collected for prospective studies or national stroke registries. To date, these data are collected manually by human readers, a labor-intensive task that is prone to errors. Purpose To evaluate the use of the large language models (LLMs) GPT-4 and GPT-3.5 to extract data from neuroradiology reports on mechanical thrombectomy in patients with ischemic stroke. Materials and Methods This retrospective study included consecutive reports from patients with ischemic stroke who underwent mechanical thrombectomy between November 2022 and September 2023 at institution 1 and between September 2016 and December 2019 at institution 2. A set of 20 reports was used to optimize the prompt, and the ability of the LLMs to extract procedural data from the reports was compared using the McNemar test. Data manually extracted by an interventional neuroradiologist served as the reference standard. Results A total of 100 internal reports from 100 patients (mean age, 74.7 years ± 13.2 [SD]; 53 female) and 30 external reports from 30 patients (mean age, 72.7 years ± 13.5; 18 male) were included. All reports were successfully processed by GPT-4 and GPT-3.5. Of 2800 data entries, 2631 (94.0% [95% CI: 93.0, 94.8]; range per category, 61%-100%) data points were correctly extracted by GPT-4 without the need for further postprocessing. With 1788 of 2800 correct data entries, GPT-3.5 produced fewer correct data entries than did GPT-4 (63.9% [95% CI: 62.0, 65.6]; range per category, 14%-99%; P < .001). For the external reports, GPT-4 extracted 760 of 840 (90.5% [95% CI: 88.3, 92.4]) correct data entries, while GPT-3.5 extracted 539 of 840 (64.2% [95% CI: 60.8, 67.4]; P < .001). Conclusion Compared with GPT-3.5, GPT-4 more frequently extracted correct procedural data from free-text reports on mechanical thrombectomy performed in patients with ischemic stroke. © RSNA, 2024 Supplemental material is available for this article.

A versatile look-up algorithm for mapping pH values and magnesium ion content using 31P MRSI.

31P MRSI allows for the non-invasive mapping of pH and magnesium ion content (Mg) in vivo, by translating the chemical shifts of inorganic phosphate and adenosine-5'-triphosphate (ATP) to pH and Mg via suitable calibration equations, such as the modified Henderson-Hasselbalch equation. However, the required constants in these calibration equations are typically only determined for physiological conditions, posing a particular challenge for their application to diseased tissue, where the biochemical conditions might change manyfold. In this article, we propose a multi-parametric look-up algorithm aiming at the condition-independent determination of pH and Mg by employing multiple quantifiable 31P spectral properties simultaneously. To generate entries for an initial look-up table, measurements from 114 model solutions prepared with varying chemical properties were made at 9.4 T. The number of look-up table entries was increased by inter- and extrapolation using a multi-dimensional function developed based on the Hill equation. The assignment of biochemical parameters, that is, pH and Mg, is realized using probability distributions incorporating specific measurement uncertainties on the quantified spectral parameters, allowing for an estimation of most plausible output values. As proof of concept, we applied a version of the look-up algorithm employing only the chemical shifts of γ- and ß-ATP for the determination of pH and Mg to in vivo 3D 31P MRSI data acquired at 7 T from (i) the lower leg muscles of healthy volunteers and (ii) the brains of patients with glioblastoma. The resulting volumetric maps showed plausible values for pH and Mg, partly revealing differences from maps generated using the conventional calibration equations.

Impact of an AI software on the diagnostic performance and reading time for the detection of cerebral aneurysms on time of flight MR-angiography.

PURPOSE: To evaluate the impact of an AI-based software trained to detect cerebral aneurysms on TOF-MRA on the diagnostic performance and reading times across readers with varying experience levels. METHODS: One hundred eighty-six MRI studies were reviewed by six readers to detect cerebral aneurysms. Initially, readings were assisted by the CNN-based software mdbrain. After 6 weeks, a second reading was conducted without software assistance. The results were compared to the consensus reading of two neuroradiological specialists and sensitivity (lesion and patient level), specificity (patient level), and false positives per case were calculated for the group of all readers, for the subgroup of physicians, and for each individual reader. Also, reading times for each reader were measured. RESULTS: The dataset contained 54 aneurysms. The readers had no experience (three medical students), 2 years experience (resident in neuroradiology), 6 years experience (radiologist), and 12 years (neuroradiologist). Significant improvements of overall specificity and the overall number of false positives per case were observed in the reading with AI support. For the physicians, we found significant improvements of sensitivity on lesion and patient level and false positives per case. Four readers experienced reduced reading times with the software, while two encountered increased times. CONCLUSION: In the reading with the AI-based software, we observed significant improvements in terms of specificity and false positives per case for the group of all readers and significant improvements of sensitivity and false positives per case for the physicians. Further studies are needed to investigate the effects of the AI-based software in a prospective setting.

Stenting with dual-layer CGuard stent in acute sub-occlusive carotid artery stenosis and in tandem occlusions: a monocentric study.

PURPOSE: Double-layer design carotid stents have been cast in a negative light since several investigations reported high rates of in-stent occlusions, at least in the acute setting of tandem occlusions. CGuard is a new generation double-layered stent that was designed to prevent periinterventional embolic events. The aim of this study was to analyze the safety and efficacy of the CGuard in emergent CAS and for the acute treatment of tandem occlusions in comparison with the single-layer Carotid Wallstent (CWS) system. METHODS: All patients who underwent CAS with CGuard or CWS after intracranial mechanical thrombectomy (MT) between 11/2018 and 12/2022 were identified from our local thrombectomy registry. Clinical, interventional and neuroimaging data were analyzed. Patency of the stent was assessed within 72 h. Intracranial hemorrhage and modified Rankin score (mRS) at discharge were the main endpoints. RESULTS: In total, 86 stent procedures in 86 patients were included (CWS: 44, CGuard: 42). CGuard had a lower, but not statistically significant rate (p = 0.431) of in-stent occlusions (n = 2, 4.8%) when compared to the CWS (n = 4, 9.1%). Significant in-stent stenosis was found in one case in each group. There was no statistically significant difference in functional outcome at discharge between the two groups with a median mRS for CGuard of 2 (IQR:1-5) vs. CWS 3 (IQR:2-4). CONCLUSION: In our series, the rate of in-stent occlusions after emergent CAS was lower with the dual-layer CGuard when compared to the monolayer CWS. Further data are needed to evaluate the potential benefit of the design in more detail.

External evaluation of a deep learning-based approach for automated brain volumetry in patients with huntington's disease.

A crucial step in the clinical adaptation of an AI-based tool is an external, independent validation. The aim of this study was to investigate brain atrophy in patients with confirmed, progressed Huntington's disease using a certified software for automated volumetry and to compare the results with the manual measurement methods used in clinical practice as well as volume calculations of the caudate nuclei based on manual segmentations. Twenty-two patients were included retrospectively, consisting of eleven patients with Huntington's disease and caudate nucleus atrophy and an age- and sex-matched control group. To quantify caudate head atrophy, the frontal horn width to intercaudate distance ratio and the intercaudate distance to inner table width ratio were obtained. The software mdbrain was used for automated volumetry. Manually measured ratios and automatically measured volumes of the groups were compared using two-sample t-tests. Pearson correlation analyses were performed. The relative difference between automatically and manually determined volumes of the caudate nuclei was calculated. Both ratios were significantly different between the groups. The automatically and manually determined volumes of the caudate nuclei showed a high level of agreement with a mean relative discrepancy of - 2.3 ± 5.5%. The Huntington's disease group showed significantly lower volumes in a variety of supratentorial brain structures. The highest degree of atrophy was shown for the caudate nucleus, putamen, and pallidum (all p < .0001). The caudate nucleus volume and the ratios were found to be strongly correlated in both groups. In conclusion, in patients with progressed Huntington's disease, it was shown that the automatically determined caudate nucleus volume correlates strongly with measured ratios commonly used in clinical practice. Both methods allowed clear differentiation between groups in this collective. The software additionally allows radiologists to more objectively assess the involvement of a variety of brain structures that are less accessible to standard semiquantitative methods.

Artificial T1-Weighted Postcontrast Brain MRI: A Deep Learning Method for Contrast Signal Extraction.

OBJECTIVES: Reducing gadolinium-based contrast agents to lower costs, the environmental impact of gadolinium-containing wastewater, and patient exposure is still an unresolved issue. Published methods have never been compared. The purpose of this study was to compare the performance of 2 reimplemented state-of-the-art deep learning methods (settings A and B) and a proposed method for contrast signal extraction (setting C) to synthesize artificial T1-weighted full-dose images from corresponding noncontrast and low-dose images. MATERIALS AND METHODS: In this prospective study, 213 participants received magnetic resonance imaging of the brain between August and October 2021 including low-dose (0.02 mmol/kg) and full-dose images (0.1 mmol/kg). Fifty participants were randomly set aside as test set before training (mean age ± SD, 52.6 ± 15.3 years; 30 men). Artificial and true full-dose images were compared using a reader-based study. Two readers noted all false-positive lesions and scored the overall interchangeability in regard to the clinical conclusion. Using a 5-point Likert scale (0 being the worst), they scored the contrast enhancement of each lesion and its conformity to the respective reference in the true image. RESULTS: The average counts of false-positives per participant were 0.33 ± 0.93, 0.07 ± 0.33, and 0.05 ± 0.22 for settings A-C, respectively. Setting C showed a significantly higher proportion of scans scored as fully or mostly interchangeable (70/100) than settings A (40/100, P < 0.001) and B (57/100, P < 0.001), and generated the smallest mean enhancement reduction of scored lesions (-0.50 ± 0.55) compared with the true images (setting A: -1.10 ± 0.98; setting B: -0.91 ± 0.67, both P < 0.001). The average scores of conformity of the lesion were 1.75 ± 1.07, 2.19 ± 1.04, and 2.48 ± 0.91 for settings A-C, respectively, with significant differences among all settings (all P < 0.001). CONCLUSIONS: The proposed method for contrast signal extraction showed significant improvements in synthesizing postcontrast images. A relevant proportion of images showing inadequate interchangeability with the reference remains at this dosage.

Association between occlusion location, net water uptake and ischemic lesion growth in large vessel anterior circulation strokes.

Ischemic lesion net water uptake (NWU) represents a quantitative imaging biomarker for cerebral edema in acute ischemic stroke. Data on NWU for distinct occlusion locations remain scarce, but might help to improve the prognostic value of NWU. In this retrospective multicenter cohort study, we compared NWU between patients with proximal large vessel occlusion (pLVO; ICA or proximal M1) and distal large vessel occlusion (dLVO; distal M1 or M2). NWU was quantified by densitometric measurements of the early ischemic region. Arterial collateral status was assessed using the Maas scale. Regression analysis was used to investigate the relationship between occlusion location, NWU and ischemic lesion growth. A total of 685 patients met inclusion criteria. Early ischemic lesion NWU was higher in patients with pLVO compared with dLVO (7.7% vs 3.9%, P < .001). The relationship between occlusion location and NWU was partially mediated by arterial collateral status. NWU was associated with absolute ischemic lesion growth between admission and follow-up imaging (ß estimate, 5.50, 95% CI, 3.81-7.19, P < .001). This study establishes a framework for the relationship between occlusion location, arterial collateral status, early ischemic lesion NWU and ischemic lesion growth. Future prognostic thresholds for NWU might be optimized by adjusting for the occlusion location.

Whole-Brain Intracellular pH Mapping of Gliomas Using High-Resolution 31P MR Spectroscopic Imaging at 7.0 T.

Malignant tumors commonly exhibit a reversed pH gradient compared with normal tissue, with a more acidic extracellular pH and an alkaline intracellular pH (pHi). In this prospective study, pHi values in gliomas were quantified using high-resolution phosphorous 31 (31P) spectroscopic MRI at 7.0 T and were used to correlate pHi alterations with histopathologic findings. A total of 12 participants (mean age, 58 years ± 18 [SD]; seven male, five female) with histopathologically proven, newly diagnosed glioma were included between September 2018 and November 2019. The 31P spectroscopic MRI scans were acquired using a double-resonant 31P/1H phased-array head coil together with a three-dimensional (3D) 31P chemical shift imaging sequence (5.7-mL voxel volume) performed with a 7.0-T whole-body system. The 3D volumetric segmentations were performed for the whole-tumor volumes (WTVs); tumor subcompartments of necrosis, gadolinium enhancement, and nonenhancing T2 (NCE T2) hyperintensity; and normal-appearing white matter (NAWM), and pHi values were compared. Spearman correlation was used to assess association between pHi and the proliferation index Ki-67. For all study participants, mean pHi values were higher in the WTV (7.057 ± 0.024) compared with NAWM (7.006 ± 0.012; P < .001). In eight participants with high-grade gliomas, pHi was increased in all tumor subcompartments (necrosis, 7.075 ± 0.033; gadolinium enhancement, 7.075 ± 0.024; NCE T2 hyperintensity, 7.043 ± 0.015) compared with NAWM (7.004 ± 0.014; all P < .01). The pHi values of WTV positively correlated with Ki-67 (R2 = 0.74, r = 0.78, P = .001). In conclusion, 31P spectroscopic MRI at 7.0 T enabled high-resolution quantification of pHi in gliomas, with pHi alteration associated with the Ki-67 proliferation index, and may aid in diagnosis and treatment monitoring. Keywords: 31P MRSI, pH, Glioma, Glioblastoma, Ultra-High-Field MRI, Imaging Biomarker, 7 Tesla Supplemental material is available for this article. © RSNA, 2023.

Synchronous versus metachronous spinal metastasis: a comparative study of survival outcomes following neurosurgical treatment.

PURPOSE: Patients with spinal metastases (SM) from solid neoplasms typically exhibit progression to an advanced cancer stage. Such metastases can either develop concurrently with an existing cancer diagnosis (termed metachronous SM) or emerge as the initial indication of an undiagnosed malignancy (referred to as synchronous SM). The present study investigates the prognostic implications of synchronous compared to metachronous SM following surgical resection. METHODS: From 2015 to 2020, a total of 211 individuals underwent surgical intervention for SM at our neuro-oncology facility. We conducted a survival analysis starting from the date of the neurosurgical procedure, comparing those diagnosed with synchronous SM against those with metachronous SM. RESULTS: The predominant primary tumor types included lung cancer (23%), prostate cancer (21%), and breast cancer (11.3%). Of the participants, 97 (46%) had synchronous SM, while 114 (54%) had metachronous SM. The median overall survival post-surgery for those with synchronous SM was 13.5 months (95% confidence interval (CI) 6.1-15.8) compared to 13 months (95% CI 7.7-14.2) for those with metachronous SM (p = 0.74). CONCLUSIONS: Our findings suggest that the timing of SM diagnosis (synchronous versus metachronous) does not significantly affect survival outcomes following neurosurgical treatment for SM. These results support the consideration of neurosurgical procedures regardless of the temporal pattern of SM manifestation.

A New Fibrin-Heparine Coated Self-Expanding Stent for the Rescue Treatment of Intracranial Stenosis-a Multicentric Study.

INTRODUCTION: Rescue intracranial stenting is necessary to provide sufficient recanalization after mechanical thrombectomy (MT) in patients with acute large vessel occlusions (LVO) due to an underlying intracranial atherosclerotic disease (ICAD). The CREDO heal is a novel stent that provides a potentially lower thrombogenicity due to surface modification. We present the first multicentric experience with the CREDO heal for acute rescue stenting. METHODS: Data of 81 patients who underwent rescue stenting after MT at 12 centers in Germany and Spain were prospectively collected and retrospectively evaluated. RESULTS: Final mTICI 2b­3 was reached in 95.1% after median two MT maneuvers and stenting. Four periprocedural complications resulted in clinical deterioration (4.9%). Intraparenchymal hemorrhage occurred in one patient (1.2%) and functional independence at FU was reached by 42% of the patients. Most interventions were performed under Gp IIb/IIIa inhibitors. CONCLUSION: CREDO heal was effective and safe in our case series. However, more data is needed to define the optimal antithrombotic regime. The use under single antiplatelet medication is not supported by our study.

Machine Learning-Based Perivascular Space Volumetry in Alzheimer Disease.

OBJECTIVES: Impaired perivascular clearance has been suggested as a contributing factor to the pathogenesis of Alzheimer disease (AD). However, it remains unresolved when the anatomy of the perivascular space (PVS) is altered during AD progression. Therefore, this study investigates the association between PVS volume and AD progression in cognitively unimpaired (CU) individuals, both with and without subjective cognitive decline (SCD), and in those clinically diagnosed with mild cognitive impairment (MCI) or mild AD. MATERIALS AND METHODS: A convolutional neural network was trained using manually corrected, filter-based segmentations (n = 1000) to automatically segment the PVS in the centrum semiovale from interpolated, coronal T2-weighted magnetic resonance imaging scans (n = 894). These scans were sourced from the national German Center for Neurodegenerative Diseases Longitudinal Cognitive Impairment and Dementia Study. Convolutional neural network-based segmentations and those performed by a human rater were compared in terms of segmentation volume, identified PVS clusters, as well as Dice score. The comparison revealed good segmentation quality (Pearson correlation coefficient r = 0.70 with P < 0.0001 for PVS volume, detection rate in cluster analysis = 84.3%, and Dice score = 59.0%). Subsequent multivariate linear regression analysis, adjusted for participants' age, was performed to correlate PVS volume with clinical diagnoses, disease progression, cerebrospinal fluid biomarkers, lifestyle factors, and cognitive function. Cognitive function was assessed using the Mini-Mental State Examination, the Comprehensive Neuropsychological Test Battery, and the Cognitive Subscale of the 13-Item Alzheimer's Disease Assessment Scale. RESULTS: Multivariate analysis, adjusted for age, revealed that participants with AD and MCI, but not those with SCD, had significantly higher PVS volumes compared with CU participants without SCD ( P = 0.001 for each group). Furthermore, CU participants who developed incident MCI within 4.5 years after the baseline assessment showed significantly higher PVS volumes at baseline compared with those who did not progress to MCI ( P = 0.03). Cognitive function was negatively correlated with PVS volume across all participant groups ( P ≤ 0.005 for each). No significant correlation was found between PVS volume and any of the following parameters: cerebrospinal fluid biomarkers, sleep quality, body mass index, nicotine consumption, or alcohol abuse. CONCLUSIONS: The very early changes of PVS volume may suggest that alterations in PVS function are involved in the pathophysiology of AD. Overall, the volumetric assessment of centrum semiovale PVS represents a very early imaging biomarker for AD.