Agreement of intracranial vessel diameters measured on 2D and 3D digital subtraction angiography using an automatic windowing algorithm

CTA-based deep-learning integrated model for identifying irregular shape and aneurysm size of unruptured intracranial aneurysms

BACKGROUND

Artificial intelligence can help to identify irregular shapes and sizes, crucial for managing unruptured intracranial aneurysms (UIAs). However, existing artificial intelligence tools lack reliable classification of UIA shape irregularity and validation against gold-standard three-dimensional rotational angiography (3DRA). This study aimed to develop and validate a deep-learning model using computed tomography angiography (CTA) for classifying irregular shapes and measuring UIA size.

METHODS

CTA and 3DRA of UIA patients from a referral hospital were included as a derivation set, with images from multiple medical centers as an external test set. Senior investigators manually measured irregular shape and aneurysm size on 3DRA as the ground truth. Convolutional neural network (CNN) models were employed to develop the CTA-based model for irregular shape classification and size measurement. Model performance for UIA size and irregular shape classification was evaluated by intraclass correlation coefficient (ICC) and area under the curve (AUC), respectively. Junior clinicians' performance in irregular shape classification was compared before and after using the model.

RESULTS

The derivation set included CTA images from 307 patients with 365 UIAs. The test set included 305 patients with 350 UIAs. The AUC for irregular shape classification of this model in the test set was 0.87, and the ICC of aneurysm size measurement was 0.92, compared with 3DRA. With the model's help, junior clinicians' performance for irregular shape classification was significantly improved (AUC 0.86 before vs 0.97 after, P<0.001).

CONCLUSION

This study provided a deep-learning model based on CTA for irregular shape classification and size measurement of UIAs with high accuracy and external validity. The model can be used to improve reader performance.

Inter-modality correlation across invasive and noninvasive angiography in the three-dimensional assessment of cerebral aneurysms

Background

Non-invasive and invasive methods of cerebral angiography, including computed tomography angiography (CTA), magnetic resonance angiography (MRA), and digital subtraction angiography (DSA), are commonly used to characterize and follow cerebral aneurysms. Prior work has validated two-dimensional size measurements across these modalities. Our study aims to compare the reliability of three-dimensional (3D) shape measurements across CTA, MRA, and DSA.

Methods

A subset of cerebral aneurysms in which more than one form of angiography was performed was selected. Aneurysms were included if they did not change in size or shape between angiographic studies. Aneurysm domes were segmented, and morphometric features were measured consistent with prior reports. Intraclass correlation coefficients (ICCs) for each morphometric measure were calculated using a two-way mixed effect model.

Results

A total of 65 individual aneurysms from 55 patients were included in the study. The majority of aneurysms were imaged with DSA and CTA (43%) or MRA and CTA (40%), with 14% having DSA, MRA, and CTA available for review. The majority of aneurysms were located in the anterior circulation (77%), with an average size was 5 (4-8) mm. ICC ranged from 0.66 to 0.99 for 3D morphometric features, corresponding to "moderate" to "excellent" correlation. Sphericity and non-sphericity index showed the lowest ICC values. With the exception of these two variables, 3D morphometrics showed "good" or "excellent" reliability. No significant difference in mean absolute difference was noted across imaging modalities for each morphometric feature.

Conclusion

The majority of 3D morphometric measures show "good" to "excellent" reliability across CTA, MRA, and DSA, allowing for comparison across imaging modalities.

Integrated Deep Learning Model for the Detection, Segmentation, and Morphologic Analysis of Intracranial Aneurysms Using CT Angiography

Purpose To develop a deep learning model for the morphologic measurement of unruptured intracranial aneurysms (UIAs) based on CT angiography (CTA) data and validate its performance using a multicenter dataset. Materials and Methods In this retrospective study, patients with CTA examinations, including those with and without UIAs, in a tertiary referral hospital from February 2018 to February 2021 were included as the training dataset. Patients with UIAs who underwent CTA at multiple centers between April 2021 and December 2022 were included as the multicenter external testing set. An integrated deep learning (IDL) model was developed for UIA detection, segmentation, and morphologic measurement using an nnU-Net algorithm. Model performance was evaluated using the Dice similarity coefficient (DSC) and intraclass correlation coefficient (ICC), with measurements by senior radiologists serving as the reference standard. The ability of the IDL model to improve performance of junior radiologists in measuring morphologic UIA features was assessed. Results The study included 1182 patients with UIAs and 578 controls without UIAs as the training dataset (median age, 55 years [IQR, 47-62 years], 1012 [57.5%] female) and 535 patients with UIAs as the multicenter external testing set (median age, 57 years [IQR, 50-63 years], 353 [66.0%] female). The IDL model achieved 97% accuracy in detecting UIAs and achieved a DSC of 0.90 (95% CI: 0.88, 0.92) for UIA segmentation. Model-based morphologic measurements showed good agreement with reference standard measurements (all ICCs > 0.85). Within the multicenter external testing set, the IDL model also showed agreement with reference standard measurements (all ICCs > 0.80). Junior radiologists assisted by the IDL model showed significantly improved performance in measuring UIA size (ICC improved from 0.88 [95% CI: 0.80, 0.92] to 0.96 [95% CI: 0.92, 0.97], P < .001). Conclusion The developed integrated deep learning model using CTA data showed good performance in UIA detection, segmentation, and morphologic measurement and may be used to assist less experienced radiologists in morphologic analysis of UIAs. Keywords: Segmentation, CT Angiography, Head/Neck, Aneurysms, Comparative Studies Supplemental material is available for this article. © RSNA, 2024 See also the commentary by Wang in this issue.

Switch Strategy from Direct Aspiration First Pass Technique to Solumbra Improves Technical Outcome in Endovascularly Treated Stroke

BACKGROUND

The major endovascular mechanic thrombectomy (MT) techniques are: Stent-Retriever (SR), aspiration first pass technique (ADAPT) and Solumbra (Aspiration + SR), which are interchangeable (defined as switching strategy (SS)). The purpose of this study is to report the added value of switching from ADAPT to Solumbra in unsuccessful revascularization stroke patients.

METHODS

This is a retrospective, single center, pragmatic, cohort study. From December 2017 to November 2019, 935 consecutive patients were admitted to the Stroke Unit and 176/935 (18.8%) were eligible for MT. In 135/176 (76.7%) patients, ADAPT was used as the first-line strategy. SS was defined as the difference between first technique adopted and the final technique. Revascularization was evaluated with modified Thrombolysis In Cerebral Infarction (TICI) with success defined as mTICI ≥ 2b. Procedural time (PT) and time to reperfusion (TTR) were recorded.

RESULTS

Stroke involved: Anterior circulation in 121/135 (89.6%) patients and posterior circulation in 14/135 (10.4%) patients. ADAPT was the most common first-line technique vs. both SR and Solumbra (135/176 (76.7%) vs. 10/176 (5.7%) vs. 31/176 (17.6%), respectively). In 28/135 (20.7%) patients, the mTICI was ≤ 2a requiring switch to Solumbra. The vessel's diameter positively predicted SS result (odd ratio (OR) 1.12, confidence of interval (CI) 95% 1.03-1.22; p = 0.006). The mean number of passes before SS was 2.0 ± 1.2. ADAPT to Solumbra improved successful revascularization by 13.3% (107/135 (79.3%) vs. 125/135 (92.6%)). PT was superior for SS comparing with ADAPT (71.1 min (CI 95% 53.2-109.0) vs. 40.0 min (CI 95% 35.0-45.2); p = 0.0004), although, TTR was similar (324.1 min (CI 95% 311.4-387.0) vs. 311.4 min (CI 95% 285.5-338.7); p = 0.23).

CONCLUSION

Successful revascularization was improved by 13.3% after switching form ADAPT to Solumbra (final mTICI ≥ 2b was 92.6%). Vessel's diameter positively predicted recourse to SS.

Tailored Vessel-Catheter Diameter Ratio in a Direct Aspiration First-Pass Technique: Is It a Matter of Caliber?

BACKGROUND AND PURPOSE

The aspiration technique has gained a prominent role in mechanical thrombectomy. The thrombectomy goal is successful revascularization (modified TICI ≥ 2b) and first-pass effect. The purpose of this study was to evaluate the impact of the vessel-catheter ratio on the modified TICI ≥ 2b and first-pass effect.

MATERIALS AND METHODS

This was a retrospective, single-center, cohort study. From January 2018 to April 2020, 111/206 (53.9%) were eligible after applying the exclusion criteria. Culprit vessel diameters were measured by 2 neuroradiologists, and the intraclass correlation coefficient was calculated. The receiver operating characteristic curve was used for assessing the vessel-catheter ratio cutoff for modified TICI ≥ 2b and the first-pass effect. Time to groin puncture and fibrinolysis were weighted using logistic regression. All possible intervals (interval size, 0.1; sliding interval, 0.01) of the vessel-catheter ratio were plotted, and the best and worst intervals were compared using the χ² test.

RESULTS

Modified TICI ≥ 2b outcome was achieved in 75/111 (67.5%), and first-pass effect was achieved in 53/75 (70.6%). The MCA diameter was 2.1 mm with an intraclass correlation coefficient of 0.92. The optimal vessel-catheter ratio cutoffs for modified TICI ≥ 2b were ≤1.51 (accuracy = 0.67; 95% CI, 0.58-0.76; P = 0.001), and for first-pass effect, they were significant (≤1.33; P = .31). The modified TICI ≥ 2b odds ratio and relative risk were 9.2 (95% CI, 2.4-36.2; P = 0.002) and 3.2 (95% CI, 1.2-8.7; P = .024). The odds ratio remained significant after logistic regression (7.4; 95% CI, 1.7-32.5; P = .008). First-pass effect odds ratio and relative risk were not significant (2.1 and 1.5; P > .05, respectively). The modified TICI ≥ 2b best and worst vessel-catheter ratio intervals were not significantly different (55.6% versus 85.7%, P = .12). The first-pass effect best vessel-catheter ratio interval was significantly higher compared with the worst one (78.6% versus 40.0%, P = .03).

CONCLUSIONS

The aspiration catheter should be selected according to culprit vessel diameter. The optimal vessel-catheter ratio cutoffs were ≤1.51 for modified TICI ≥ 2b with an odds ratio of 9.2 and a relative risk of 3.2.