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.

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.

Sustainability in interventional radiology: are we doing enough to save the environment?

BACKGROUND: Healthcare waste contributes substantially to the world's carbon footprint. Our aims are to review the current knowledge of Interventional Radiology (IR) waste generation and ways of reducing waste in practice, to quantify the environmental and financial impact of waste generated and address green initiatives to improve IR waste management. METHODS: A systematic literature search was conducted in July 2022 using the Medline and Embase literature databases. The scope of the search included the field of IR as well as operating theatre literature, where relevant to IR practice. RESULTS: One-hundred articles were reviewed and 68 studies met the inclusion criteria. Greening initiatives include reducing, reusing and recycling waste, as well as strict waste segregation. Interventional radiologists can engage with suppliers to reformulate procedure packs to minimize unnecessary items and packaging. Opened but unused equipment can be prevented if there is better communication within the team and increased staff awareness of wasted equipment cost. Incentives to use soon-to-expire equipment can be offered. Power consumption can be reduced by powering down operating room lights and workstations when not in use, changing to Light Emitting Diode (LED) and motion sensor lightings. Surgical hand wash can be replaced with alcohol-based hand rubs to reduce water usage. Common barriers to improving waste management include the lack of leadership, misconceptions regarding infectious risk, lack of data, concerns about increased workload, negative staff attitudes and resistance to change. Education remains a top priority to engage all staff in sustainable healthcare practices. CONCLUSION: Interventional radiologists have a crucial role to play in improving healthcare sustainability. By implementing small, iterative changes to our practice, financial savings, greater efficiency and improved environmental sustainability can be achieved.

Intracerebral hemorrhage markers on non-contrast computed tomography as predictors of the dynamic spot sign on CT perfusion and associations with hematoma expansion and outcome.

PURPOSE: To assess the association between non-contrast computed tomography (NCCT) hematoma markers and the dynamic spot sign on computed tomography perfusion (CTP), and their associations with hematoma expansion, clinical outcome, and in-hospital mortality. METHODS: Patients who presented with intracerebral hemorrhage (ICH) to a stroke center over an 18-month period and underwent baseline NCCT and CTP, and a follow-up NCCT within 24 h after the baseline scan were included. The initial and follow-up hematoma volumes were calculated. Two raters independently assessed the baseline NCCT for hematoma markers and concurrently assessed the CTP for the dynamic spot sign. Univariate and multivariate logistic regression analyses were performed to assess the association between the hematoma markers and the dynamic spot sign, adjusting for known ICH expansion predictors. RESULTS: Eighty-five patients were included in our study and 55 patients were suitable for expansion analysis. Heterogeneous density was the only NCCT hematoma marker to be associated with the dynamic spot sign after multivariate analysis (odds ratio, 58.61; 95% confidence interval, 9.13-376.05; P < 0.001). The dynamic spot sign was present in 22 patients (26%) and significantly predicted hematoma expansion (odds ratio, 36.6; 95% confidence interval, 2.51-534.2; P = 0.008). All patients with a spot sign had a swirl sign. A co-located hypodensity and spot sign was significantly associated with in-hospital mortality (odds ratio, 6.17; 95% confidence interval, 1.09-34.78; P = 0.039). CONCLUSION: Heterogeneous density and swirl sign are associated with the dynamic spot sign. The dynamic spot sign is a stronger predictor than NCCT hematoma markers of significant hematoma expansion. A co-located spot sign and hypodensity predicts in-hospital mortality.

Intracranial atherosclerotic disease and acute ischaemic stroke: A review of diagnosis and management.

Intracranial atherosclerotic disease is highly prevalent and a common cause of ischaemic stroke globally. With the increasing use of endovascular treatment for acute stroke management, computed tomography and magnetic resonance imaging have become an essential part of patient selection. In this review, we present the typical imaging findings of intracranial atherosclerosis and an overview of management as relevant to diagnostic and interventional radiologists.

Development of a machine learning-based real-time location system to streamline acute endovascular intervention in acute stroke: a proof-of-concept study.

BACKGROUND: Delivery of acute stroke endovascular intervention can be challenging because it requires complex coordination of patient and staff across many different locations. In this proof-of-concept paper we (a) examine whether WiFi fingerprinting is a feasible machine learning (ML)-based real-time location system (RTLS) technology that can provide accurate real-time location information within a hospital setting, and (b) hypothesize its potential application in streamlining acute stroke endovascular intervention. METHODS: We conducted our study in a comprehensive stroke care unit in Melbourne, Australia that offers a 24-hour mechanical thrombectomy service. ML algorithms including K-nearest neighbors, decision tree, random forest, support vector machine and ensemble models were trained and tested on a public WiFi dataset and the study hospital WiFi dataset. The hospital dataset was collected using the WiFi explorer software (version 3.0.2) on a MacBook Pro (AirPort Extreme, Broadcom BCM43x×1.0). Data analysis was implemented in the Python programming environment using the scikit-learn package. The primary statistical measure for algorithm performance was the accuracy of location prediction. RESULTS: ML-based WiFi fingerprinting can accurately predict the different hospital zones relevant in the acute endovascular intervention workflow such as emergency department, CT room and angiography suite. The most accurate algorithms were random forest and support vector machine, both of which were 98% accurate. The algorithms remain robust when new data points, which were distinct from the training dataset, were tested. CONCLUSIONS: ML-based RTLS technology using WiFi fingerprinting has the potential to streamline delivery of acute stroke endovascular intervention by efficiently tracking patient and staff movement during stroke calls.

CaRotid Artery Filtering Technique (CRAFT): A Technique for Carotid Artery Stenting with Intrinsic Embolic Protection.

PURPOSE: Carotid artery stenting (CAS) is an established treatment for symptomatic carotid artery stenosis as an alternative to carotid endarterectomy. A variety of techniques and devices have been devised to minimise periprocedural stroke risk using either proximal or distal embolic protection. This study presents a method of embolic protection during CAS-the CaRotid Artery Filtering Technique (CRAFT). MATERIALS AND METHODS: The CRAFT technique employs aspects of both proximal and distal embolic protection. The CASPER RX stent (MicroVention, Tustin, CA, USA), which is a double-layered, closed-cell, micromesh nitinol stent, is deployed across the carotid artery stenosis with the assistance of a FlowGate balloon guide catheter (Stryker Neurovascular, Fremont, CA, USA). The partially deployed stent acts as a distal filter while the balloon guide is deflated midway during stent deployment to prevent distal plaque embolisation, followed by completion of stent deployment and angioplasty. RESULTS: A total of 94 patients underwent CAS using the CRAFT technique between June 2016 and March 2021. Successful stent deployment was achieved in all patients. Preliminary results demonstrated acute stent occlusion in 6 patients (6.4%) and distal embolic stroke in 5 patients (5.3%). The median procedural fluoroscopy time was 34 minutes with an interquartile range of 22 to 55 minutes. CONCLUSION: The CRAFT technique of CAS presented by this study can be applied in the treatment of symptomatic carotid artery stenosis in both emergency and elective procedure settings with a high technical success and low distal embolic stroke risk.

Radiologic Mimics of Osteomyelitis and Septic Arthritis: A Pictorial Essay.

Various imaging techniques may be employed in the investigation of suspected bone and joint infections. These include ultrasound, radiography, functional imaging such as positron emission tomography (PET) and nuclear scintigraphy, and cross-sectional imaging, including computed tomography (CT) and magnetic resonance imaging (MRI). The cross-sectional modalities represent the imaging workhorse in routine practice. The role of imaging also extends to include assessment of the anatomical extent of infection, potentially associated complications, and treatment response. The imaging appearances of bone and joint infections are heterogeneous and depend on the duration of infection, an individual patient's immune status, and virulence of culprit organisms. To add to the complexity of radiodiagnosis, one of the pitfalls of imaging musculoskeletal infection is the presence of other conditions that can share overlapping imaging features. This includes osteoarthritis, vasculopathy, inflammatory, and even neoplastic processes. Different pathologies may also coexist, for example, diabetic neuropathy and osteomyelitis. This pictorial review aims to highlight potential mimics of osteomyelitis and septic arthritis that are regularly encountered, with emphasis on specific imaging features that may aid the radiologist and clinician in distinguishing an infective from a noninfective aetiology.

Endovascular balloon-assisted liquid embolisation of soft tissue vascular malformations: technical feasibility and safety.

PURPOSE: Arteriovenous malformations (AVMs) are abnormal communications between arteries and veins without an intervening capillary system. The best endovascular treatment option for these is unclear and may involve multiple staged procedures using a variety of embolic materials. We report our initial experience using a modified version of a previously published neurointerventional technique to treat soft tissue AVMs with single-stage curative intent. MATERIALS AND METHODS: Soft tissue AVMs treated endovascularly using either sole arterial or combined arterial and venous balloon-assisted techniques with liquid embolic agents were retrospectively identified over a 3.5 year period (January 2017 to June 2020)) at two centres. Clinical, pre-operative radiological, procedural technical and post treatment details were recorded. RESULTS: Seven patients were treated for symptomatic soft tissue arteriovenous malformations. These AVMs were located in the peripheral limbs (five), tongue (one) and uterus (one). Curative treatment was achieved in 6/7 patients with one patient requiring a second treatment approximately 1 year later. A variety of liquid embolisation agents (LEAs) including sclerosants and polymers were used. Clinical success rate was 100% following treatment. One patient experienced expected temporary post-operative tongue swelling requiring tracheostomy occurred following embolisation of the lingual AVM. A minor complication in a second patient was due to an access site haematoma developed following treatment of the hand AVM requiring surgical intervention. No long-term sequelae or additional complications were observed. CONCLUSION: Endovascular arterial and venous balloon assisted LEA embolization of soft tissue AVMs with curative intent is feasible. This technique may provide an alternative treatment option for achieving durable occlusion for complex soft tissue AVMs.

Application of a computational model in simulating an endovascular clot retrieval service system within regional Australia.

INTRODUCTION: The global demand for endovascular clot retrieval (ECR) has grown rapidly in recent years creating challenges to healthcare system planning and resource allocation. This study aims to apply our established computational model to predict and optimise the performance and resource allocation of ECR services within regional Australia, and applying data from the state of South Australia as a modelling exercise. METHOD: Local geographic information obtained using the Google Maps application program interface and real-world data was input into the discrete event simulation model we previously developed. The results were obtained after the simulation was run over 5 years. We modelled and compared a single-centre and two-centre ECR service delivery system. RESULTS: Based on the input data, this model was able to simulate the ECR delivery system in the state of South Australia from the moment when emergency services were notified of a potential stroke patient to potential delivery of ECR treatment. In the model, ECR delivery improved using a two-centre system compared to a one-centre system, as the percentage of stroke patients requiring ECR was increased. When 15% of patients required ECR, the proportion of 'failure to receive ECR' cases for a single-centre system was 17.35%, compared to 3.71% for a two-centre system. CONCLUSIONS: Geolocation and resource utilisation within the ECR delivery system are crucial in optimising service delivery and patient outcome. Under the model assumptions, as the number of stroke cases requiring ECR increased, a two-centre ECR system resulted in increased timely ECR delivery, compared to a single-centre system. This study demonstrated the flexibility and the potential application of our DES model in simulating the stroke service within any location worldwide.

Artificial intelligence in clinical decision support and outcome prediction - applications in stroke.

Artificial intelligence (AI) is making a profound impact in healthcare, with the number of AI applications in medicine increasing substantially over the past five years. In acute stroke, it is playing an increasingly important role in clinical decision-making. Contemporary advances have increased the amount of information - both clinical and radiological - which clinicians must consider when managing patients. In the time-critical setting of acute stroke, AI offers the tools to rapidly evaluate and consolidate available information, extracting specific predictions from rich, noisy data. It has been applied to the automatic detection of stroke lesions on imaging and can guide treatment decisions through the prediction of tissue outcomes and long-term functional outcomes. This review examines the current state of AI applications in stroke, exploring their potential to reform stroke care through clinical decision support, as well as the challenges and limitations which must be addressed to facilitate their acceptance and adoption for clinical use.

Antiplatelet Drugs for Neurointerventions: Part 2 Clinical Applications.

Endovascular techniques have expanded to include balloon and stent-assistance, flow diversion and individualized endovascular occlusion devices, to widen the treatment spectrum for more complex aneurysm morphologies. While usually well-tolerated by patients, endovascular treatment of intracranial aneurysms carries the risk of complications, with procedure-related ischemic complications being the most common. Several antiplatelet agents have been studied in a neurointerventional setting for both prophylaxis and in the setting of intraprocedural thrombotic complications. Knowledge of these antiplatelet agents, evidence for their use and common dosages is important for the practicing neurointerventionist to ensure the proper application of these agents.Part one of this two-part review focused on basic platelet physiology, pharmacology of common antiplatelet medications and future directions and therapies. Part two focuses on clinical applications and evidence based therapeutic regimens.

Fronto-parietal involvement in chronic stroke motor performance when corticospinal tract integrity is compromised.

BACKGROUND: Preserved integrity of the corticospinal tract (CST) is a marker of good upper-limb behavior and recovery following stroke. However, there is less understanding of neural mechanisms that might help facilitate upper-limb motor recovery in stroke survivors with extensive CST damage. OBJECTIVE: The purpose of this study was to investigate resting state functional connectivity in chronic stroke survivors with different levels of CST damage and to explore neural correlates of greater upper-limb motor performance in stroke survivors with compromised ipsilesional CST integrity. METHODS: Thirty chronic stroke survivors (24 males, aged 64.7 ± 10.8 years) participated in this study. Three experimental sessions were conducted to: 1) obtain anatomical (T1, T2) structural (diffusion) and functional (resting state) MRI sequences, 2) determine CST integrity with transcranial magnetic stimulation (TMS) and conduct assessments of upper-limb behavior, and 3) reconfirm CST integrity status. Participants were divided into groups according to the extent of CST damage. Those in the extensive CST damage group did not show TMS evoked responses and had significantly lower ipsilesional fractional anisotropy. RESULTS: Of the 30 chronic stroke survivors, 12 were categorized as having extensive CST damage. Stroke survivors with extensive CST damage had weaker functional connectivity in the ipsilesional sensorimotor network and greater functional connectivity in the ipsilesional fronto-parietal network compared to those with preserved CST integrity. For participants with extensive CST damage, improved motor performance was associated with greater functional connectivity of the ipsilesional fronto-parietal network and higher fractional anisotropy of the ipsilesional rostral superior longitudinal fasciculus. CONCLUSIONS: Stroke survivors with extensive CST damage have greater resting state functional connectivity of an ipsilesional fronto-parietal network that appears to be a behaviorally relevant neural mechanism that improves upper-limb motor performance.

Review of deep learning algorithms for the automatic detection of intracranial hemorrhages on computed tomography head imaging.

Artificial intelligence is a rapidly evolving field, with modern technological advances and the growth of electronic health data opening new possibilities in diagnostic radiology. In recent years, the performance of deep learning (DL) algorithms on various medical image tasks have continually improved. DL algorithms have been proposed as a tool to detect various forms of intracranial hemorrhage on non-contrast computed tomography (NCCT) of the head. In subtle, acute cases, the capacity for DL algorithm image interpretation support might improve the diagnostic yield of CT for detection of this time-critical condition, potentially expediting treatment where appropriate and improving patient outcomes. However, there are multiple challenges to DL algorithm implementation, such as the relative scarcity of labeled datasets, the difficulties in developing algorithms capable of volumetric medical image analysis, and the complex practicalities of deployment into clinical practice. This review examines the literature and the approaches taken in the development of DL algorithms for the detection of intracranial hemorrhage on NCCT head studies. Considerations in crafting such algorithms will be discussed, as well as challenges which must be overcome to ensure effective, dependable implementations as automated tools in a clinical setting.

Tranexamic acid in patients with intracerebral haemorrhage (STOP-AUST): a multicentre, randomised, placebo-controlled, phase 2 trial.

BACKGROUND: Despite intracerebral haemorrhage causing 5% of deaths worldwide, few evidence-based therapeutic strategies other than stroke unit care exist. Tranexamic acid decreases haemorrhage in conditions such as acute trauma and menorrhoea. We aimed to assess whether tranexamic acid reduces intracerebral haemorrhage growth in patients with acute intracerebral haemorrhage. METHODS: We did a prospective, double-blind, randomised, placebo-controlled, investigator-led, phase 2 trial at 13 stroke centres in Australia, Finland, and Taiwan. Patients were eligible if they were aged 18 years or older, had an acute intracerebral haemorrhage fulfilling clinical criteria (eg, Glasgow Coma Scale score of >7, intracerebral haemorrhage volume <70 mL, no identified or suspected secondary cause of intracerebral haemorrhage, no thrombotic events within the previous 12 months, no planned surgery in the next 24 h, and no use of anticoagulation), had contrast extravasation on CT angiography (the so-called spot sign), and were treatable within 4·5 h of symptom onset and within 1 h of CT angiography. Patients were randomly assigned (1:1) to receive either 1 g of intravenous tranexamic acid over 10 min followed by 1 g over 8 h or matching placebo, started within 4·5 h of symptom onset. Randomisation was done using a centralised web-based procedure with randomly permuted blocks of varying size. All patients, investigators, and staff involved in patient management were masked to treatment. The primary outcome was intracerebral haemorrhage growth (>33% relative or >6 mL absolute) at 24 h. The primary and safety analyses were done in the intention-to-treat population. The trial is registered at ClinicalTrials.gov (NCT01702636). FINDINGS: Between March 1, 2013, and Aug 13, 2019, we enrolled and randomly assigned 100 participants to the tranexamic acid group (n=50) or the placebo group (n=50). Median age was 71 years (IQR 57-79) and median intracerebral haemorrhage volume was 14·6 mL (7·9-32·7) at baseline. The primary outcome was not different between the two groups: 26 (52%) patients in the placebo group and 22 (44%) in the tranexamic acid group had intracerebral haemorrhage growth (odds ratio [OR] 0·72 [95% CI 0·32-1·59], p=0·41). There was no evidence of a difference in the proportions of patients who died or had thromboembolic complications between the groups: eight (16%) in the placebo group vs 13 (26%) in the tranexamic acid group died and two (4%) vs one (2%) had thromboembolic complications. None of the deaths was considered related to study medication. INTERPRETATION: Our study does not provide evidence that tranexamic acid prevents intracerebral haemorrhage growth, although the treatment was safe with no increase in thromboembolic complications. Larger trials of tranexamic acid, with simpler recruitment methods and an earlier treatment window, are justified. FUNDING: National Health and Medical Research Council, Royal Melbourne Hospital Foundation.

Environmental sustainability in neurointerventional procedures: a waste audit.

BACKGROUND: Operating rooms contribute between 20% to 70% of hospital waste. This study aimed to evaluate the waste burden of neurointerventional procedures performed in a radiology department, identify areas for waste reduction, and motivate new greening initiatives. METHODS: We performed a waste audit of 17 neurointerventional procedures at a tertiary-referral center over a 3-month period. Waste was categorized into five streams: general waste, clinical waste, recyclable plastic, recyclable paper, and sharps. Our radiology department started recycling soft plastics from 13 December 2019. Hence, an additional recyclable soft plastic waste stream was added from this time point. The weight of each waste stream was measured using a digital weighing scale. RESULTS: We measured the waste from seven cerebral digital subtraction angiograms (DSA), six mechanical thrombectomies (MT), two aneurysm-coiling procedures, one coiling with tumour embolization, and one dural arteriovenous fistula embolization procedure. In total, the 17 procedures generated 135.3 kg of waste: 85.5 kg (63.2%) clinical waste, 28.0 kg (20.7%) general waste, 14.7 kg (10.9%) recyclable paper, 3.5 kg (2.6%) recyclable plastic, 2.2 kg (1.6%) recyclable soft plastic, and 1.4 kg (1.0%) of sharps. An average of 8 kg of waste was generated per case. Coiling cases produced the greatest waste burden (13.1 kg), followed by embolization (10.3 kg), MT (8.8 kg), and DSA procedures (5.1 kg). CONCLUSION: Neurointerventional procedures generate a substantial amount of waste, an average of 8 kg per case. Targeted initiatives such as engaging with suppliers to revise procedure packs and reduce packaging, digitizing paper instructions, opening devices only when necessary, implementing additional recycling programs, and appropriate waste segregation have the potential to reduce the environmental impact of our specialty.

Antiplatelet Drugs for Neurointerventions: Part 1 Clinical Pharmacology.

The development of endovascular treatment for intracranial aneurysms has established new techniques such as balloon and stent-assistance, flow diversion and endosaccular occlusion devices. Antiplatelet treatment is an important aspect to reduce risk of thrombus formation on microcatheters and implanted devices when utilizing these methods. It is particularly relevant for flow diverting stents to prevent early and late stent thrombosis. Consideration of platelet physiology and appropriate selection of antiplatelet medication is important as platelet dysfunction drives many of the pathological processes and complications of neurointerventional procedures. Part one of this review focuses on basic platelet physiology, pharmacology of common antiplatelet medications and future directions and therapies. Part two focuses on clinical applications and evidence-based therapeutic regimens.

Geographic Service Delivery for Endovascular Clot Retrieval: Using Discrete Event Simulation to Optimize Resources.

BACKGROUND: Endovascular clot retrieval (ECR) is the standard of care for acute ischemic stroke caused by large vessel occlusion. Reducing stroke symptom onset to reperfusion time is associated with improved functional outcomes. This study aims to develop a computational model to predict and identify time-related outcomes of community stroke calls within a geographic area based on variable parameters to support planning and coordination of ECR services. METHODS: A discrete event simulation (DES) model to simulate and predict ECR service was designed using SimPy, a process-based DES framework written in Python. Geolocation data defined by the user, as well as that used by the model, were sourced using the Google Maps application programming interface. Variables were customized by the user on the basis of their local environment to provide more accurate prediction. RESULTS: A DES model can estimate the delay between the time that emergency services are notified of a potential stroke and potential cerebral reperfusion using ECR at a capable hospital. Variables can be adjusted to observe the effect of modifying each parameter input. By varying the percentage of stroke patients receiving ECR, we were able to define the levels at which our existing service begins to fail in service delivery and assess the effect of adding centers. CONCLUSIONS: This novel computational DES model can aid the optimization of delivery of a stroke service within a city, state, or country. By varying geographic, population, and other user-defined inputs, the model can be applied to any location worldwide.