Clinical Neuroradiology最新文献

Purpose: GPT‑4 has been shown to correctly extract procedural details from free-text reports on mechanical thrombectomy. However, GPT may not be suitable for analyzing reports containing personal data. The purpose of this study was to evaluate the ability of the large language models (LLM) Llama3.1 405B, Llama3 70B, Llama3 8B, and Mixtral 8X7B, that can be operated offline, to extract procedural details from free-text reports on mechanical thrombectomies.

Methods: Free-text reports on mechanical thrombectomy from two institutions were included. A detailed prompt was used in German and English languages. The ability of the LLMs to extract procedural data was compared to GPT‑4 using McNemar's test. The manual data entries made by an interventional neuroradiologist served as the reference standard.

Results: 100 reports from institution 1 (mean age 74.7 ± 13.2 years; 53 females) and 30 reports from institution 2 (mean age 72.7 ± 13.5 years; 18 males) were included. Llama 3.1 405B extracted 2619 of 2800 data points correctly (93.5% [95%CI: 92.6%, 94.4%], p = 0.39 vs. GPT-4). Llama3 70B with the English prompt extracted 2537 data points correctly (90.6% [95%CI: 89.5%, 91.7%], p < 0.001 vs. GPT-4), and 2471 (88.2% [95%CI: 87.0%, 89.4%], p < 0.001 vs. GPT-4) with the German prompt. Llama 3 8B extracted 2314 data points correctly (86.1% [95%CI: 84.8%, 87.4%], p < 0.001 vs. GPT-4), and Mixtral 8X7B extracted 2411 (86.1% [95%CI: 84.8%, 87.4%], p < 0.001 vs. GPT-4) correctly.

Conclusion: Llama 3.1 405B was equal to GPT‑4 for data extraction from free-text reports on mechanical thrombectomies and may represent a data secure alternative, when operated locally.

Purpose: Assess the prevalence of thalamic AVM (tAVMs), their revealing conditions and angio-architectural features, as well as their therapeutic management.

Methods: Monocentric retrospective study (1998 to 2018) involving 748 consecutive patients with a brain AVM managed at the Pitié-Salpêtrière Hospital, from which tAVMs were retrieved. Revealing condition of the tAVMs was recorded. tAVMs' angio-architecture was systematically reviewed on angiographic examinations. The safety and effectiveness of treatments (conservative management and exclusion treatment) were evaluated based on clinical and angiographic criteria.

Results: Twenty-two consecutive patients (9 males, mean age: 32 ± 14 years) with a tAVM were included (prevalence: 2.9%). tAVMs were revealed by a bleeding event in 86.4% of cases. The posterolateral choroidal artery was the most frequently observed arterial feeder in our series (71.4%). Associated aneurysm (nidal and/or pedicular) was observed in 36.4% of patients. The tAVMs' mean volume was 6.45 ± 12.8 ml. Venous drainage was in the deep venous system only in most cases (81%). Four patients received conservative management, 12 patients were treated by radiosurgery and 12 by embolization, with 6 patients (27.2%) who received both embolization and radiosurgery. Complete nidal occlusion was achieved in 33.3% of the patients who underwent angiographic follow-up. Two embolized patients (16.7%) suffered severe embolization-related complications, and 4 patients treated with radiosurgery (33.3%) suffered severe procedure related complication, all of which resolved without significant sequela.

Conclusion: Our results underline the safety of exclusion treatments by embolization and/or radiosurgery of tAVMs. Efficacy in terms of angiographic exclusion still needs to be demonstrated at long-term due to limited imaging follow-up.

Background and purpose: Incomplete reperfusion in endovascular thrombectomy (EVT) impacts patients' outcomes. Different incomplete reperfusion patterns may benefit from targeted therapeutic strategies, e.g. EVT-accessible incomplete reperfusion patterns could improve by performing additional EVT attempts, while EVT-non-accessible incomplete patterns might benefit from pharmacological therapies. The health-economic implications of these therapies are uncertain. This study aims to assess the potential economic benefits of improving incomplete reperfusion patterns after EVT.

Materials and methods: Retrospective Data analysis from the ESCAPE-NA1 trial, which included patients with large vessel occlusion strokes undergoing EVT. Reperfusion patterns were classified as near-/complete (eTICI 2c3), EVT-accessible incomplete (eTICI 2b), or EVT-non-accessible incomplete (eTICI 2b) and we compared multiple attempts to achieve eTICI 2c3 vs. first-pass eTICI 2c3. A Markov-Model was built to compare lifetime costs and quality adjusted life-years (QALY) for each reperfusion pattern over a lifetime horizon, considering both healthcare and societal perspectives.

Results: A total of 1105 of patients were enrolled in the ESCAPE-NA1 trial of which 949 with eTICI 2b, 2c and 3 were further analyized (mean age 70.7 ± 13.6 [SD]; 463 female). Near-Complete reperfusion (eTICI 2c3) was achieved in 506/1105 patients (45.8%). Incomplete reperfusion patterns (eTICI 2b) were found in 450/1105 (40.7%) patients. Angiography imaging could be further investigated in 443/450 (98.4%) cases with 147/443(33.2%) EVT-accessible and 296/443(66.8%) EVT-non-accessible incomplete reperfusion patterns. Compared to EVT-accessible and EVT-non-accssible incomplete reperfusion, achieving complete (eTICI 2c3) reperfusion resulted in lower costs and an additional 1.14/0.45 QALYs, making it the dominant strategy from a health-economic perspective. In the complete reperfusion (eTICI 2c3) group, cumulative lifetime QALYs were similar with 5.25 for single-pass eTICI 2c3 and 5.19 for multi-pass eTICI 2c3.

Conclusions: Improving incomplete reperfusion patterns after EVT has considerable potential health economic benefits, both in the presence and absence of a target occlusion that is amenable to EVT.

Introduction/background: The rising demand for endovascular stroke therapy highlights the need for standardized training environments. Studies have shown the positive impact of simulator-based training for neurointerventional procedures. Given the frequent use of percutaneous transluminal angioplasty (PTA) of the internal carotid artery (ICA) in acute settings, specialized simulation training is crucial. This study evaluates the feasibility of a custom-made ICA stenosis model for PTA simulation.

Methods: Internally developed 3D-printed circular clamps were attached to patient-based 3D printed ICA models and integrated into a previously described realistic whole-body neurovascular simulation model HANNES (Hamburg Anatomic Neurointerventional Endovascular Simulator) to simulate a proximal ICA stenosis. Participants (N = 5) of varying experience levels each performed three PTA procedures. Fluoroscopy time, radiation doses, and the extent of stenosis and balloon inflation pressure were assessed. After simulation, participants rated the model in terms of haptic, feasibility and applicability.

Results: For statistical analysis, the participants were divided into two groups according to their experience level. There was a statistically significant difference in the procedure duration between both groups, U = 10.500, Z = -1.968, p < 0.05. Significant test results could also be demonstrated for radiation doses (DAP) among groups, Kolmogorov-Smirnov p < 0.05, U = 7.000, Z = -2.357, p < 0.05. No difference was shown between the total contrast volume (ml) among groups that was used during procedures (m = 11.67, SD 2.09).

Conclusion: The authors propose a novel ICA stenosis simulation model for training of the cervical PTA. The model provides a realistic and replicable method for standardized procedural training.