Using angiographic parametric imaging-derived radiomics features to predict complications and embolization outcomes of intracranial aneurysms treated by pipeline embolization devices

Predictors of aneurysm occlusion after treatment with flow diverters: a systematic literature review

BACKGROUND

Flow diverters (FDs) have become an integral part of treatment for brain aneurysms.

AIM

To summarize available evidence of factors associated with aneurysm occlusion (AO) after treatment with a FD.

METHODS

References were identified using the Nested Knowledge AutoLit semi-automated review platform between January 1, 2008 and August 26, 2022. The review focuses on preprocedural and postprocedural factors associated with AO identified in logistic regression analysis. Studies were included if they met the inclusion criteria of study details (ie, study design, sample size, location, (pre)treatment aneurysm details). Evidence levels were classified by variability and significancy across studies (eg, low variability ≥5 studies and significance in ≥60% throughout reports).

RESULTS

Overall, 2.03% (95% CI 1.22 to 2.82; 24/1184) of screened studies met the inclusion criteria for predictors of AO based on logistic regression analysis. Predictors of AO with low variability in multivariable logistic regression analysis included aneurysm characteristics (aneurysm diameter), particularly complexity (absence of branch involvement) and younger patient age. Predictors of moderate evidence for AO included aneurysm characteristics (neck width), patient characteristics (absence of hypertension), procedural (adjunctive coiling) and post-deployment variables (longer follow-up; direct postprocedural satisfactory occlusion). Variables with a high variability in predicting AO following FD treatment were gender, FD as re-treatment strategy, and aneurysm morphology (eg, fusiform or blister).

CONCLUSION

Evidence of predictors for AO after FD treatment is sparse. Current literature suggests that absence of branch involvement, younger age, and aneurysm diameter have the highest impact on AO following FD treatment. Large studies investigating high-quality data with well-defined inclusion criteria are needed for greater insight into FD effectiveness.

Advances in research and application of artificial intelligence and radiomic predictive models based on intracranial aneurysm images

Intracranial aneurysm is a high-risk disease, with imaging playing a crucial role in their diagnosis and treatment. The rapid advancement of artificial intelligence in imaging technology holds promise for the development of AI-based radiomics predictive models. These models could potentially enable the automatic detection and diagnosis of intracranial aneurysms, assess their status, and predict outcomes, thereby assisting in the creation of personalized treatment plans. In addition, these techniques could improve diagnostic efficiency for physicians and patient prognoses. This article aims to review the progress of artificial intelligence radiomics in the study of intracranial aneurysms, addressing the challenges faced and future prospects, in hopes of introducing new ideas for the precise diagnosis and treatment of intracranial aneurysms.

Prediction of obliteration of unruptured cerebral aneurysm by residual aneurysm volume after flow diverter stent treatment

Introduction There is no methodology to predict aneurysm occlusion using residual volume after flow diverter stent treatment. We retrospectively examined whether residual aneurysm volume at 6 months postoperatively can predict the degree of aneurysm obliteration at 1 year after flow diverter stent treatment. Materials and Methods This single institution study included 101 consecutive patients who underwent flow diverter stent treatment for unruptured cerebral aneurysm. Based on pre-treatment aneurysm volume, the percentage residual volume was calculated 6 months postoperatively. The volume of the aneurysm was determined using the volume calculation function of the cerebral angiography equipment. 1 year postoperatively, patients were classified into two groups: the good obliteration group (GG; O'KellyMarotta [OKM] grading scale: C and D) and the poor obliteration group (PG; OKM: A and B). Statistical analysis was performed to determine if there was a difference in residual aneurysm volume percentage at 6 months postoperatively between the two groups. Results A total of 20 patients were studied: 6 in the GG and 14 in the PG. Mean residual aneurysm volume at 6 months postoperatively in the GG was 33.1% (±34.7), while that in the PG was 80.6% (±24.8) (P=0.018). A residual aneurysm volume of ≥35.2% at 6 months postoperatively was significantly associated with poor aneurysm obliteration at 1 year postoperatively (AUC=0.88, P=0.008). Conclusions Residual aneurysm volume percentage at 6 months after flow diverter stent treatment might be able to predict the likelihood of aneurysm occlusion at 1 year postoperatively.

Predicting postinterventional rupture of intracranial aneurysms using arteriography-derived radiomic features after pipeline embolization

Background and purpose

Postinterventional rupture of intracranial aneurysms (IAs) remains a severe complication after flow diverter treatment. However, potential hemodynamic mechanisms underlying independent predictors for postinterventional rupture of IAs remain unclear. In this study, we employed arteriography-derived radiomic features to predict this complication.

Methods

We included 64 patients who underwent pipeline flow diversion for intracranial aneurysms, distinguishing between 16 patients who experienced postinterventional rupture and 48 who did not. We performed propensity score matching based on clinical and morphological factors to match these patients with 48 patients with postinterventional unruptured IAs at a 1:3 ratio. Postinterventional digital subtraction angiography were used to create five arteriography-derived perfusion parameter maps and then radiomics features were obtained from each map. Informative features were selected through the least absolute shrinkage and selection operator method with five-fold cross-validation. Subsequently, radiomics scores were formulated to predict the occurrence of postinterventional IA ruptures. Prediction performance was evaluated with the training and test datasets using area under the curve (AUC) and confusion matrix-derived metrics.

Results

Overall, 1,459 radiomics features were obtained, and six were selected. The resulting radiomics scores had high efficacy in distinguishing the postinterventional rupture group. The AUC and Youden index were 0.912 (95% confidence interval: 0.767-1.000) and 0.847 for the training dataset, respectively, and 0.938 (95% confidence interval, 0.806-1.000) and 0.800 for the testing dataset, respectively.

Conclusion

Radiomics scores generated using arteriography-derived radiomic features effectively predicted postinterventional IA ruptures and may aid in differentiating IAs at high risk of postinterventional rupture.

High-resolution magnetic resonance imaging-based radiomic features aid in selecting endovascular candidates among patients with cerebral venous sinus thrombosis

OBJECTIVES

Cerebral venous sinus thrombosis (CVST) can cause sinus obstruction and stenosis, with potentially fatal consequences. High-resolution magnetic resonance imaging (HRMRI) can diagnose CVST qualitatively, although quantitative screening methods are lacking for patients refractory to anticoagulation therapy and who may benefit from endovascular treatment (EVT). Thus, in this study, we used radiomic features (RFs) extracted from HRMRI to build machine learning models to predict response to drug therapy and determine the appropriateness of EVT.

MATERIALS AND METHODS

RFs were extracted from three-dimensional T1-weighted motion-sensitized driven equilibrium (MSDE), T2-weighted MSDE, T1-contrast, and T1-contrast MSDE sequences to build radiomic signatures and support vector machine (SVM) models for predicting the efficacy of standard drug therapy and the necessity of EVT.

RESULTS

We retrospectively included 53 patients with CVST in a prospective cohort study, among whom 14 underwent EVT after standard drug therapy failed. Thirteen RFs were selected to construct the RF signature and CVST-SVM models. In the validation dataset, the sensitivity, specificity, and area under the curve performance for the RF signature model were 0.833, 0.937, and 0.977, respectively. The radiomic score was correlated with days from symptom onset, history of dyslipidemia, smoking, fibrin degradation product, and D-dimer levels. The sensitivity, specificity, and area under the curve for the CVST-SVM model in the validation set were 0.917, 0.969, and 0.992, respectively.

CONCLUSIONS

The CVST-SVM model trained with RFs extracted from HRMRI outperformed the RF signature model and could aid physicians in predicting patient responses to drug treatment and identifying those who may require EVT.

Drawing time-density curve with Fiji/ImageJ: An alternative approach for parametric coding of cerebral digital subtraction angiography

BACKGROUND

Quantitative spatiotemporal analysis of digital subtraction angiography could support clinical decision making for the management of cerebral vascular disease. However, there is a lack of free and user-friendly applications. The objective of our study is to devise a free and simple solution for parametric coding of digital subtraction angiography.

NEW METHOD

By driving the time-density curves in the region of interest, the digital subtraction angiography images were color-coded and quantitatively analyzed using fully open-source and free software (Fiji/ImageJ). The similarity factor (f2) was used to compare the resolution profiles between time-density curves generated with commercial software on the Siemens workstation (syngo iFlow, Siemens Healthcare, Berlin, Germany) and our method.

RESULTS AND COMPARISON WITH EXISTING METHOD

Sixteen patients diagnosed with acute ischemic stroke resulting from acute occlusion of the distal internal carotid artery or the first segment of the middle cerebral artery were selected for analysis. Angiography images were successfully processed with syngo iFlow and Fiji/ImageJ. The images processed with Fiji/ImageJ provided excellent anatomic and hemodynamic details. In all patients, the similarity factor (f2) values of the time-density curves derived from the same region of interest were 99.90 (range 99.85-99.95).

CONCLUSIONS

The ImageJ/Fiji software provides a user-friendly and free alternative for parametric coding of digital subtraction angiography.

Quantitative evaluation of the subsequent hemorrhage with arteriography-derived hemodynamic features in patients with untreated cerebral arteriovenous malformation

Background

Patients with untreated cerebral arteriovenous malformations (AVMs) are at risk of intracerebral hemorrhage. However, treatment to prevent AVM hemorrhage carries risks.

Objective

This study aimed to analyze the AVM nidus-related hemodynamic features and identify the risk factors for subsequent hemorrhage.

Methods

We retrospectively identified patients with untreated AVMs who were assessed at our institution between March 2010 and March 2021. Patients with ≥6 months of treatment-free and hemorrhage-free follow-up after diagnosed by digital subtraction angiography were included in subsequent examinations. The hemodynamic features were extracted from five contrast flow-related parameter maps. The Kaplan-Meier analyses and Cox proportional hazards regression models were used to find the potential risk factors for subsequent hemorrhage.

Results

Overall, 104 patients with a mean follow-up duration of 3.37 years (median, 2.42 years; range, 6-117 months) were included in study, and the annual risk of rupture was 3.7%. Previous rupture (hazard ratio [HR], 4.89; 95% confidence interval [CI], 1.16-20.72), deep AVM location (HR, 4.02; 95% CI, 1.01-15.99), higher cerebral blood volume (HR, 3.35; 95% CI, 1.15-9.74) in the nidus, and higher stasis index (HR, 1.54; 95% CI, 1.06-2.24) in the nidus were associated with subsequent hemorrhage in untreated AVMs.

Conclusion

Higher cerebral blood volume and stasis index in the nidus suggest increased blood inflow and stagnant blood drainage. The combination of these factors may cause subsequent hemorrhage of AVMs.

Prediction of the trans-stenotic pressure gradient with arteriography-derived hemodynamic features in patients with idiopathic intracranial hypertension

The pathogenesis of idiopathic intracranial hypertension (IIH) is attributed to segmental stenosis of the venous sinus. The current treatment paradigm requires a trans-stenotic pressure gradient of ≥8 mmHg or ≥6 mmHg threshold. This study aimed to develop a machine learning screening method to identify patients with IIH using hemodynamic features. A total of 204 venous manometry instances (n = 142, training and validation; n = 62, test) from 135 patients were included. Radiomic features extracted from five arteriography perfusion parameter maps were selected using least absolute shrinkage and selection operator and then entered into support vector machine (SVM) classifiers. The Thr8-23-SVM classifier was created with 23 radiomic features to predict if the pressure gradient was ≥8 mmHg. On an independent test dataset, prediction sensitivity, specificity, accuracy, and AUC were 0.972, 0.846, 0.919, and 0.980, respectively (95% confidence interval: 0.980-1.000). For the 6 mmHg threshold, thr6-28-SVM incorporated 28 features, and its sensitivity, specificity, accuracy, and AUC were 0.923, 0.956, 0.935, and 0.969, respectively (95% confidence interval: 0.927-1.000). The trans-stenotic pressure gradient result was associated with perfusion pattern changes, and SVM classifiers trained with arteriography perfusion map-derived radiomic features could predict the 8 mmHg and 6 mmHg dichotomized trans-stenotic pressure gradients with favorable accuracy.