Tranexamic acid as a novel adjunct in the management of vessel perforation complicating Endovascular Clot Retrieval.

AIM: To assess the safety and utility of tranexamic acid (TXA) as an adjunct salvage therapy in iatrogenic vessel perforation complicating endovascular clot retrieval. Iatrogenic vessel perforation and extravasation are known and potentially fatal complications of endovascular clot retrieval (ECR). Various methods of establishing haemostasis post perforation have been reported. TXA is widely utilised intraoperatively to reduce bleeding in various surgical specialities. The use of TXA in endovascular procedures has not been previously described in the literature. METHODS: Retrospective case control study of all cases that underwent ECR. Cases where arterial rupture occurred were identified. Details of management and functional status at 3 months were recorded. Modified Rankin score (mRS) 0-2 was considered a good functional outcome. Comparison of proportions analysis was performed. RESULTS: Of 1378 cases of ECR, rupture complicated 36 (2.6%). TXA was administered in addition to standard care in 11 cases (31%). At 3 months, 4 of 11 cases (36%) where TXA was administered had a good functional outcome compared to 3 of 22 (12%) in the standard care group (P=0.09). Mortality at 3 months occurred in 4 of 11 cases (41.7%) where TXA was administered compared to 16 of 25 (64%) where it was not (P=0.13). CONCLUSION: Tranexamic acid administration in iatrogenic vessel rupture was associated with a lower mortality rate and a larger proportion of patients achieving a good functional outcome at 3 months. This effect trended towards but was not statistically significant. TXA administration was not associated with adverse effects.

Tenecteplase versus Alteplase before Thrombectomy for Ischemic Stroke.

BACKGROUND: Intravenous infusion of alteplase is used for thrombolysis before endovascular thrombectomy for ischemic stroke. Tenecteplase, which is more fibrin-specific and has longer activity than alteplase, is given as a bolus and may increase the incidence of vascular reperfusion. METHODS: We randomly assigned patients with ischemic stroke who had occlusion of the internal carotid, basilar, or middle cerebral artery and who were eligible to undergo thrombectomy to receive tenecteplase (at a dose of 0.25 mg per kilogram of body weight; maximum dose, 25 mg) or alteplase (at a dose of 0.9 mg per kilogram; maximum dose, 90 mg) within 4.5 hours after symptom onset. The primary outcome was reperfusion of greater than 50% of the involved ischemic territory or an absence of retrievable thrombus at the time of the initial angiographic assessment. Noninferiority of tenecteplase was tested, followed by superiority. Secondary outcomes included the modified Rankin scale score (on a scale from 0 [no neurologic deficit] to 6 [death]) at 90 days. Safety outcomes were death and symptomatic intracerebral hemorrhage. RESULTS: Of 202 patients enrolled, 101 were assigned to receive tenecteplase and 101 to receive alteplase. The primary outcome occurred in 22% of the patients treated with tenecteplase versus 10% of those treated with alteplase (incidence difference, 12 percentage points; 95% confidence interval [CI], 2 to 21; incidence ratio, 2.2; 95% CI, 1.1 to 4.4; P=0.002 for noninferiority; P=0.03 for superiority). Tenecteplase resulted in a better 90-day functional outcome than alteplase (median modified Rankin scale score, 2 vs. 3; common odds ratio, 1.7; 95% CI, 1.0 to 2.8; P=0.04). Symptomatic intracerebral hemorrhage occurred in 1% of the patients in each group. CONCLUSIONS: Tenecteplase before thrombectomy was associated with a higher incidence of reperfusion and better functional outcome than alteplase among patients with ischemic stroke treated within 4.5 hours after symptom onset. (Funded by the National Health and Medical Research Council of Australia and others; EXTEND-IA TNK ClinicalTrials.gov number, NCT02388061 .).

Endovascular thrombectomy for tandem acute ischemic stroke associated with cervical artery dissection: a systematic review and meta-analysis.

PURPOSE: Strokes associated with cervical artery dissection have been managed primarily with antithrombotics with poor outcomes. The additive role of endovascular thrombectomy remains unclear. The objective was to perform systematic review and meta-analysis to compare endovascular thrombectomy and medical therapy for acute ischemic stroke associated with cervical artery dissection. METHODS: Studies from six electronic databases included outcomes of patient cohorts with acute ischemic stroke secondary to cervical artery dissection who underwent treatment with endovascular thrombectomy. A meta-analysis of proportions was conducted with a random effects model. Modified Rankin score at 90 days (mRS 0-2) was the primary outcome. Other outcomes included proportion of patients with thrombolysis in cerebral infarction (TICI) 2b-3 flow, 90-day mortality rate, and 90-day symptomatic intracerebral hemorrhage (sICH) rate. RESULTS: Six studies were included, comprising 193 cases that underwent thrombectomy compared with 59 cases that were managed medically. Successful recanalization with a pooled proportion of thrombolysis in cerebral infarction (TICI) 2b-3 flow in the thrombectomy group was 74%. Favorable outcome (mRS 0-2) was superior in the pooled thrombectomy group (62.9%, 95% CI 55.8-69.5%) compared with medical management (41.5%, 95% CI 29.0-55.1%, P = 0.006). The pooled rate of 90-day mortality was similar for endovascular vs medical (8.6% vs 6.3%). The pooled rate of symptomatic intracranial haemorrhage (sICH) did not significantly differ (5.9% vs 4.2%, P = 0.60). CONCLUSIONS: Current data suggest that endovascular thrombectomy may be an option in patients with acute ischemic stroke due to cervical artery dissection. This requires further confirmation in higher quality prospective studies.

Endovascular clot retrieval for M2 segment middle cerebral artery occlusion: a systematic review and meta-analysis.

BACKGROUND: Endovascular clot retrieval (ECR) is the standard of care for acute ischaemic stroke due to large vessel occlusion. However, isolated occlusion of the M2 segment of the middle cerebral artery (MCA) was underrepresented in the landmark trials. AIMS: Given the potential treatment benefit associated with M2 MCA occlusions, we aimed to evaluate the outcome of patients undergoing ECR for M2 occlusion. METHODS: We conducted a systematic review and meta-analysis of the available literature that included patients with M2 MCA occlusions who underwent ECR. Successful reperfusion was defined as a treatment in cerebral ischaemia score of 2b-3. Good outcome was defined as a modified Rankin Scale score ≤ 2. We also analysed complications such as post-procedure symptomatic intracranial haemorrhage and mortality at 3 months. RESULTS: Fifteen studies including 1105 patients with isolated M2 occlusions were analysed. Successful reperfusion occurred in 75.4% (95% confidence interval (CI) 67.7-84.1%) of patients; good outcome was observed in 58.3% (95% CI 51.7-63.8% of patients. The rate of symptomatic intracranial haemorrhage was 5.1% (95% CI 4.2-8.3%), and 3-month mortality rate was 12.2% (95% CI 10.4-16.3%). CONCLUSION: The outcomes of ECR treatment of M2 occlusions are favourable, with good safety profile. Comparison to medical management from large registries or randomised controlled trials is warranted.

Effect of Intravenous Tenecteplase Dose on Cerebral Reperfusion Before Thrombectomy in Patients With Large Vessel Occlusion Ischemic Stroke: The EXTEND-IA TNK Part 2 Randomized Clinical Trial.

Importance: Intravenous thrombolysis with tenecteplase improves reperfusion prior to endovascular thrombectomy for ischemic stroke compared with alteplase. Objective: To determine whether 0.40 mg/kg of tenecteplase safely improves reperfusion before endovascular thrombectomy vs 0.25 mg/kg of tenecteplase in patients with large vessel occlusion ischemic stroke. Design, Setting, and Participants: Randomized clinical trial at 27 hospitals in Australia and 1 in New Zealand using open-label treatment and blinded assessment of radiological and clinical outcomes. Patients were enrolled from December 2017 to July 2019 with follow-up until October 2019. Adult patients (N = 300) with ischemic stroke due to occlusion of the intracranial internal carotid, \basilar, or middle cerebral artery were included less than 4.5 hours after symptom onset using standard intravenous thrombolysis eligibility criteria. Interventions: Open-label tenecteplase at 0.40 mg/kg (maximum, 40 mg; n = 150) or 0.25 mg/kg (maximum, 25 mg; n = 150) given as a bolus before endovascular thrombectomy. Main Outcomes and Measures: The primary outcome was reperfusion of greater than 50% of the involved ischemic territory prior to thrombectomy, assessed by consensus of 2 blinded neuroradiologists. Prespecified secondary outcomes were level of disability at day 90 (modified Rankin Scale [mRS] score; range, 0-6); mRS score of 0 to 1 (freedom from disability) or no change from baseline at 90 days; mRS score of 0 to 2 (functional independence) or no change from baseline at 90 days; substantial neurological improvement at 3 days; symptomatic intracranial hemorrhage within 36 hours; and all-cause death. Results: All 300 patients who were randomized (mean age, 72.7 years; 141 [47%] women) completed the trial. The number of participants with greater than 50% reperfusion of the previously occluded vascular territory was 29 of 150 (19.3%) in the 0.40 mg/kg group vs 29 of 150 (19.3%) in the 0.25 mg/kg group (unadjusted risk difference, 0.0% [95% CI, -8.9% to -8.9%]; adjusted risk ratio, 1.03 [95% CI, 0.66-1.61]; P = .89). Among the 6 secondary outcomes, there were no significant differences in any of the 4 functional outcomes between the 0.40 mg/kg and 0.25 mg/kg groups nor in all-cause deaths (26 [17%] vs 22 [15%]; unadjusted risk difference, 2.7% [95% CI, -5.6% to 11.0%]) or symptomatic intracranial hemorrhage (7 [4.7%] vs 2 [1.3%]; unadjusted risk difference, 3.3% [95% CI, -0.5% to 7.2%]). Conclusions and Relevance: Among patients with large vessel occlusion ischemic stroke, a dose of 0.40 mg/kg, compared with 0.25 mg/kg, of tenecteplase did not significantly improve cerebral reperfusion prior to endovascular thrombectomy. The findings suggest that the 0.40-mg/kg dose of tenecteplase does not confer an advantage over the 0.25-mg/kg dose in patients with large vessel occlusion ischemic stroke in whom endovascular thrombectomy is planned. Trial Registration: ClinicalTrials.gov Identifier: NCT03340493.

Peering Into the Black Box of Artificial Intelligence: Evaluation Metrics of Machine Learning Methods.

OBJECTIVE: Machine learning (ML) and artificial intelligence (AI) are rapidly becoming the most talked about and controversial topics in radiology and medicine. Over the past few years, the numbers of ML- or AI-focused studies in the literature have increased almost exponentially, and ML has become a hot topic at academic and industry conferences. However, despite the increased awareness of ML as a tool, many medical professionals have a poor understanding of how ML works and how to critically appraise studies and tools that are presented to us. Thus, we present a brief overview of ML, explain the metrics used in ML and how to interpret them, and explain some of the technical jargon associated with the field so that readers with a medical background and basic knowledge of statistics can feel more comfortable when examining ML applications. CONCLUSION: Attention to sample size, overfitting, underfitting, cross validation, as well as a broad knowledge of the metrics of machine learning, can help those with little or no technical knowledge begin to assess machine learning studies. However, transparency in methods and sharing of algorithms is vital to allow clinicians to assess these tools themselves.

A Fine Needle Recanalization Technique for Iliac Artery Occlusions in Endovascular Surgery.

BACKGROUND: Endovascular intervention for chronic total occlusions (CTOs) in aortoiliac occlusive disease (AIOD) poses technical challenges. In this manuscript, our experience of fine needle recanalization for the treatment of iliac artery CTO is described. METHOD: A prospective database recorded treatment of 11 limbs in 11 patients since 2011 using this technique. The majority of these CTO were of the common iliac artery (n = 9). RESULTS: Technical success rate was 91% (n = 10). One failed case was due to tortuous iliac anatomy. There was no restenosis of the treated segments at 8 weeks and no major complications, perforations, major limb loss, or periprocedural mortality. CONCLUSIONS: This technique is a safe and viable adjunct for difficult CTO in AIOD with suitable anatomy. It benefits from being a simple, low-profile, low-cost coaxial system and should be part of the armamentarium with other advanced endovascular techniques.

Outcomes of endovascular thrombectomy with and without bridging thrombolysis for acute large vessel occlusion ischaemic stroke.

BACKGROUND: Endovascular thrombectomy (EVT) for management of large vessel occlusion (LVO) acute ischaemic stroke is now current best practice. AIM: To determine if bridging intravenous (i.v.) alteplase therapy confers any clinical benefit. METHODS: A retrospective study of patients treated with EVT for LVO was performed. Outcomes were compared between patients receiving thrombolysis and EVT with EVT alone. Primary end-points were reperfusion rate, 90-day functional outcome and mortality using the modified Rankin Scale (mRS) and symptomatic intracranial haemorrhage (sICH). RESULTS: A total of 355 patients who underwent EVT was included: 210 with thrombolysis (59%) and 145 without (41%). The reperfusion rate was higher in the group receiving i.v. tissue plasminogen activator (tPA) (unadjusted odds ratio (OR) 2.2, 95% confidence interval (CI): 1.29-3.73, P = 0.004), although this effect was attenuated when all variables were considered (adjusted OR (AOR) 1.22, 95% CI: 0.60-2.5, P = 0.580). The percentage achieving functional independence (mRS 0-2) at 90 days was higher in patients who received bridging i.v. tPA (AOR 2.17, 95% CI: 1.06-4.44, P = 0.033). There was no significant difference in major complications, including sICH (AOR 1.4, 95% CI: 0.51-3.83, P = 0.512). There was lower 90-day mortality in the bridging i.v. tPA group (AOR 0.79, 95% CI: 0.36-1.74, P = 0.551). Fewer thrombectomy passes (2 versus 3, P = 0.012) were required to achieve successful reperfusion in the i.v. tPA group. Successful reperfusion (modified thrombolysis in cerebral infarction ≥2b) was the strongest predictor for 90-day functional independence (AOR 10.4, 95% CI:3.6-29.7, P < 0.001). CONCLUSION: Our study supports the current practice of administering i.v. alteplase before endovascular therapy.

Endovascular therapy for ischemic stroke with perfusion-imaging selection.

BACKGROUND: Trials of endovascular therapy for ischemic stroke have produced variable results. We conducted this study to test whether more advanced imaging selection, recently developed devices, and earlier intervention improve outcomes. METHODS: We randomly assigned patients with ischemic stroke who were receiving 0.9 mg of alteplase per kilogram of body weight less than 4.5 hours after the onset of ischemic stroke either to undergo endovascular thrombectomy with the Solitaire FR (Flow Restoration) stent retriever or to continue receiving alteplase alone. All the patients had occlusion of the internal carotid or middle cerebral artery and evidence of salvageable brain tissue and ischemic core of less than 70 ml on computed tomographic (CT) perfusion imaging. The coprimary outcomes were reperfusion at 24 hours and early neurologic improvement (≥8-point reduction on the National Institutes of Health Stroke Scale or a score of 0 or 1 at day 3). Secondary outcomes included the functional score on the modified Rankin scale at 90 days. RESULTS: The trial was stopped early because of efficacy after 70 patients had undergone randomization (35 patients in each group). The percentage of ischemic territory that had undergone reperfusion at 24 hours was greater in the endovascular-therapy group than in the alteplase-only group (median, 100% vs. 37%; P<0.001). Endovascular therapy, initiated at a median of 210 minutes after the onset of stroke, increased early neurologic improvement at 3 days (80% vs. 37%, P=0.002) and improved the functional outcome at 90 days, with more patients achieving functional independence (score of 0 to 2 on the modified Rankin scale, 71% vs. 40%; P=0.01). There were no significant differences in rates of death or symptomatic intracerebral hemorrhage. CONCLUSIONS: In patients with ischemic stroke with a proximal cerebral arterial occlusion and salvageable tissue on CT perfusion imaging, early thrombectomy with the Solitaire FR stent retriever, as compared with alteplase alone, improved reperfusion, early neurologic recovery, and functional outcome. (Funded by the Australian National Health and Medical Research Council and others; EXTEND-IA ClinicalTrials.gov number, NCT01492725, and Australian New Zealand Clinical Trials Registry number, ACTRN12611000969965.).

Cerebral arterial gas embolism from attempted mechanical thrombectomy: recovery following hyperbaric oxygen therapy.

Cerebral arterial gas embolism is a recognised complication of endovascular intervention with an estimated incidence of 0.08%. Its diagnosis is predominantly clinical, supported by neuroimaging. The treatment relies on alleviating mechanical obstruction and reversing the proinflammatory processes that contribute to tissue ischaemia. Hyperbaric oxygen therapy is an effective treatment and has multiple mechanisms to reverse the pathological processes involved in cerebral arterial gas embolism. Symptomatic cerebral arterial gas embolism is a rare complication of endovascular intervention for acute ischaemic stroke. Although there are no previous descriptions of its successful treatment with hyperbaric oxygen therapy following mechanical thrombectomy, this is likely to become more common as mechanical thrombectomy is increasingly used worldwide to treat acute ischaemic stroke.

Structure-based discovery of LpxC inhibitors.

The emergence and spread of multidrug-resistant (MDR) Gram negative bacteria presents a serious threat for public health. Novel antimicrobials that could overcome the resistance problems are urgently needed. UDP-3-O-(R-3-hydroxymyristol)-N-acetylglucosamine deacetylase (LpxC) is a cytosolic zinc-based deacetylase that catalyzes the first committed step in the biosynthesis of lipid A, which is essential for the survival of Gram-negative bacteria. Our efforts toward the discovery of novel LpxC inhibitors are presented herein.

Crystal structure of a signal recognition particle Alu domain in the elongation arrest conformation.

The signal recognition particle (SRP) is a conserved ribonucleoprotein particle that targets membrane and secreted proteins to translocation channels in membranes. In eukaryotes, the Alu domain, which comprises the 5' and 3' extremities of the SRP RNA bound to the SRP9/14 heterodimer, is thought to interact with the ribosome to pause translation elongation during membrane docking. We present the 3.2 Å resolution crystal structure of a chimeric Alu domain, comprising Alu RNA from the archaeon Pyrococcus horikoshii bound to the human Alu binding proteins SRP9/14. The structure reveals how intricate tertiary interactions stabilize the RNA 5' domain structure and how an extra, archaeal-specific, terminal stem helps constrain the Alu RNA into the active closed conformation. In this conformation, highly conserved noncanonical base pairs allow unusually tight side-by-side packing of 5' and 3' RNA stems within the SRP9/14 RNA binding surface. The biological relevance of this structure is confirmed by showing that a reconstituted full-length chimeric archaeal-human SRP is competent to elicit elongation arrest in vitro. The structure will be useful in refining our understanding of how the SRP Alu domain interacts with the ribosome.

A novel protein-protein interaction in the RES (REtention and Splicing) complex.

The retention and splicing (RES) complex is a conserved spliceosome-associated module that was shown to enhance splicing of a subset of transcripts and promote the nuclear retention of unspliced pre-mRNAs in yeast. The heterotrimeric RES complex is organized around the Snu17p protein that binds to both the Bud13p and Pml1p subunits. Snu17p exhibits an RRM domain that resembles a U2AF homology motif (UHM) and Bud13p harbors a Trp residue reminiscent of an UHM-ligand motif (ULM). It has therefore been proposed that the interaction between Snu17p and Bud13p resembles canonical UHM-ULM complexes. Here, we have used biochemical and NMR structural analysis to characterize the structure of the yeast Snu17p-Bud13p complex. Unlike known UHMs that sequester the Trp residue of the ULM ligand in a hydrophobic pocket, Snu17p and Bud13p utilize a large interaction surface formed around the two helices of the Snu17p domain. In total 18 residues of the Bud13p ligand wrap around the Snu17p helical surface in an U-turn-like arrangement. The invariant Trp(232) in Bud13p is located in the center of the turn, and contacts surface residues of Snu17p. The structural data are supported by mutational analysis and indicate that Snu17p provides an extended binding surface with Bud13p that is notably distinct from canonical UHM-ULM interactions. Our data highlight structural diversity in RRM-protein interactions, analogous to the one seen for nucleic acid interactions.

Structure-function analysis of pneumococcal DprA protein reveals that dimerization is crucial for loading RecA recombinase onto DNA during transformation.

Transformation promotes genome plasticity in bacteria via RecA-driven homologous recombination. In the gram-positive human pathogen Streptococcus pneumoniae, the transformasome a multiprotein complex, internalizes, protects, and processes transforming DNA to generate chromosomal recombinants. Double-stranded DNA is internalized as single strands, onto which the transformation-dedicated DNA processing protein A (DprA) ensures the loading of RecA to form presynaptic filaments. We report that the structure of DprA consists of the association of a sterile alpha motif domain and a Rossmann fold and that DprA forms tail-to-tail dimers. The isolation of DprA self-interaction mutants revealed that dimerization is crucial for the formation of nucleocomplexes in vitro and for genetic transformation. Residues important for DprA-RecA interaction also were identified and mutated, establishing this interaction as equally important for transformation. Positioning of key interaction residues on the DprA structure revealed an overlap of DprA-DprA and DprA-RecA interaction surfaces. We propose a model in which RecA interaction promotes rearrangement or disruption of the DprA dimer, enabling the subsequent nucleation of RecA and its polymerization onto ssDNA.

High-frequency optical coherence tomography for endovascular management of cerebral aneurysms.

Endovascular management of intracranial aneurysms has become the mainstay of treatment in recent years; however, retreatment rates remain as high as 1 in 5. High-frequency optical coherence tomography (HF-OCT) is an emerging imaging modality for the assessment, treatment and follow-up of cerebral aneurysms. EMBASE and SCOPUS databases were searched for studies relating to the management of intracranial aneurysm with OCT. A combination of keywords were used including 'cerebral aneurysm', 'intracranial aneurysm', 'high-frequency optical coherence tomography', 'optical coherence tomography', and 'optical frequency domain imaging'. There were 23 papers included in this review. For the assessment of intracranial aneurysm, OCT was able to accurately assess aneurysm morphology as well as detailed analysis of arterial wall layers. During IA treatment, OCT was used to assess and troubleshoot stent placement to optimise successful isolation from the circulation. In the follow-up period, endothelial growth patterns were visualised by OCT imaging. OCT shows promise for the treatment of IAs at all stages of management. Due to the novel development of HF-OCT, there is limited longitudinal data in human studies. Further research in this area is required with a focus specifically on long-term treatment outcomes in humans.

Verification of a simplified aneurysm dimensionless flow parameter to predict intracranial aneurysm rupture status.

OBJECTIVES: Aneurysm number (An) is a novel prediction tool utilizing parameters of pulsatility index (PI) and aneurysm geometry. An has been shown to have the potential to differentiate intracranial aneurysm (IA) rupture status. The objective of this study is to investigate the feasibility and accuracy of An for IA rupture status prediction using Australian based clinical data. METHODS: A retrospective study was conducted across three tertiary referral hospitals between November 2017 and November 2020 and all saccular IAs with known rupture status were included. Two sets of An values were calculated based on two sets of PI values previously reported in the literature. RESULTS: Five hundred and four IA cases were included in this study. The results demonstrated no significant difference between ruptured and unruptured status when using An ≥1 as the discriminator. Further analysis showed no strong correlation between An and IA subtypes. The area under the curve (AUC) indicated poor performance in predicting rupture status (AUC1 = 0.55 and AUC2 = 0.56). CONCLUSIONS: This study does not support An ≥1 as a reliable parameter to predict the rupture status of IAs based on a retrospective cohort. Although the concept of An is supported by hemodynamic aneurysm theory, further research is needed before it can be applied in the clinical setting. ADVANCES IN KNOWLEDGE: This study demonstrates that the novel prediction tool, An, proposed in 2020 is not reliable and that further research of this hemodynamic model is needed before it can be incorporated into the prediction of IA rupture status.

The Clinical and Radiological Outcomes of the Multimodal Use of the Woven EndoBridge Device: A Large Multicenter Study.

BACKGROUND: The Woven EndoBridge (WEB) is a device used for intrasaccular flow diversion, designed for the elimination of wide-necked bifurcation aneurysms from the circulation. In this study, we aim to assess the safety and efficacy of the WEB and its uses in treating aneurysms of different morphologies and locations. METHODS: In a retrospective analysis, we compiled a comprehensive dataset from patients treated with the WEB device across three major Australian neurovascular centers from May 2017 to September 2023. The case series encompassed a spectrum of aneurysm types, including wide-necked bifurcation, sidewall, and irregularly shaped aneurysms, as well as cases previously managed with alternative therapeutic strategies. This study additionally encompasses cases where aneurysms were managed using the WEB device in combination with supplementary endovascular devices. RESULTS: The study included 169 aneurysms in 161 patients. The rate of satisfactory aneurysm occlusion was 85.6%, with 86.7% of patients maintaining good functional status at their most recent follow-up. The procedure exhibited a low mortality rate of 0.6% and a thromboembolic complication rate of 7.1% (n = 12/161). There were no instances of postoperative re-rupture and the procedure-related hemorrhage rate was low (1.2%, n = 2/169), aligning with the literature regarding the safety and efficacy of the WEB device. CONCLUSIONS: Our multicenter trial reinforces the WEB device's role as an effective and safe modality for intracranial aneurysm management, supporting its expanded application beyond wide-necked bifurcation aneurysms. Further prospective studies are required to delineate its evolving role fully.

Evaluation of techniques to improve a deep learning algorithm for the automatic detection of intracranial haemorrhage on CT head imaging.

BACKGROUND: Deep learning (DL) algorithms are playing an increasing role in automatic medical image analysis. PURPOSE: To evaluate the performance of a DL model for the automatic detection of intracranial haemorrhage and its subtypes on non-contrast CT (NCCT) head studies and to compare the effects of various preprocessing and model design implementations. METHODS: The DL algorithm was trained and externally validated on open-source, multi-centre retrospective data containing radiologist-annotated NCCT head studies. The training dataset was sourced from four research institutions across Canada, the USA and Brazil. The test dataset was sourced from a research centre in India. A convolutional neural network (CNN) was used, with its performance compared against similar models with additional implementations: (1) a recurrent neural network (RNN) attached to the CNN, (2) preprocessed CT image-windowed inputs and (3) preprocessed CT image-concatenated inputs. The area under the receiver operating characteristic curve (AUC-ROC) and microaveraged precision (mAP) score were used to evaluate and compare model performances. RESULTS: The training and test datasets contained 21,744 and 491 NCCT head studies, respectively, with 8,882 (40.8%) and 205 (41.8%) positive for intracranial haemorrhage. Implementation of preprocessing techniques and the CNN-RNN framework increased mAP from 0.77 to 0.93 and increased AUC-ROC [95% confidence intervals] from 0.854 [0.816-0.889] to 0.966 [0.951-0.980] (p-value = 3.91 × 10-12). CONCLUSIONS: The deep learning model accurately detected intracranial haemorrhage and improved in performance following specific implementation techniques, demonstrating clinical potential as a decision support tool and an automated system to improve radiologist workflow efficiency. KEY POINTS: • The deep learning model detected intracranial haemorrhages on computed tomography with high accuracy. • Image preprocessing, such as windowing, plays a large role in improving deep learning model performance. • Implementations which enable an analysis of interslice dependencies can improve deep learning model performance. • Visual saliency maps can facilitate explainable artificial intelligence systems. • Deep learning within a triage system may expedite earlier intracranial haemorrhage detection.