Evaluation of an Artificial Intelligence-Based 3D-Angiography for Visualization of Cerebral Vasculature

Fractal analysis of healthy and diseased vasculature in pediatric Moyamoya disease

BACKGROUND AND PURPOSE

Fractal dimension is an objective metric for the notion of structural complexity. We sought to investigate differences in structural complexity between healthy and affected territories of cerebral vasculature in moyamoya, as well as associated scalp vasculature and native transdural collaterals in patients with moyamoya by comparing their respective fractal dimensions.

METHODS

Our cohort consisted of 15 transdural collaterals from 12 patients with unilateral anterior circulation moyamoya. Frames of distal arterial vasculature from internal and external carotid angiograms were selected then automatically segmented and also manually annotated by a cerebrovascular surgeon. In the affected hemisphere, the region with transdural collateral supply was compared to the contralateral region. The resulting skeletonized angiograms were analyzed for their fractal dimensions.

RESULTS

We found the average fractal dimension (Df) of the moyamoya-side ICA was 1.82 with slightly different means for the anteroposterial (AP) and lateral views (mean  =  1.82; mean  =  1.81). The overall mean for healthy cerebral vasculature was also found to be 1.82 (AP: mean  =  1.83; lateral: mean  =  1.81). Mean Df of native transdural collaterals was found to be 1.82 (AP: mean  =  1.83; lateral: mean  =  1.81). The mean Df difference between autosegmented and manually segmented images was 0.013.

CONCLUSION

In accordance with the clinical understanding of moyamoya disease, the distal arterial structural complexity is not affected in moyamoya, and is maintained by transdural collaterals formed by vasculogenesis. Autosegmentation of cerebral vasculature is also shown to be accurate when compared to manual segmentation.

Calculation of virtual 3D subtraction angiographies using conditional generative adversarial networks (cGANs)

OBJECTIVE

Subtraction angiographies are calculated using a native and a contrast-enhanced 3D angiography images. This minimizes both bone and metal artifacts and results in a pure image of the vessels. However, carrying out the examination twice means double the radiation dose for the patient. With the help of generative AI, it could be possible to simulate subtraction angiographies from contrast-enhanced 3D angiographies and thus reduce the need for another dose of radiation without a cutback in quality. We implemented this concept by using conditional generative adversarial networks.

METHODS

We selected all 3D subtraction angiographies from our PACS system, which had performed between 01/01/2018 and 12/31/2022 and randomly divided them into training, validation, and test sets (66%:17%:17%). We adapted the pix2pix framework to work on 3D data and trained a conditional generative adversarial network with 621 data sets. Additionally, we used 158 data sets for validation and 164 for testing. We evaluated two test sets with (n = 72) and without artifacts (n = 92). Five (blinded) neuroradiologists compared these datasets with the original subtraction dataset. They assessed similarity, subjective image quality, and severity of artifacts.

RESULTS

Image quality and subjective diagnostic accuracy of the virtual subtraction angiographies revealed no significant differences compared to the original 3D angiographies. While bone and movement artifact level were reduced, artifact level caused by metal implants differed from case to case between both angiographies without one group being significant superior to the other.

CONCLUSION

Conditional generative adversarial networks can be used to simulate subtraction angiographies in clinical practice, however, new artifacts can also appear as a result of this technology.

Application of deblur technology for improving the clarity of digital subtractive angiography

BACKGROUND

Digital subtraction angiography (DSA) is most commonly used in vessel disease examinations and treatments. We aimed to develop a novel deep learning-based method to deblur the large focal spot DSA images, so as to obtain a clearer and sharper cerebrovascular DSA image.

METHODS

The proposed network cascaded several residual dense blocks (RDBs), which contain dense connected layers and local residual learning. Several loss functions for image restoration were investigated. Our training set consisted of 52 paired images of angiography with more than 350,000 cropped patches. The testing set included 10 body phantoms and 80 clinical images of different types of diseases for subjective evaluation. All test images were acquired using a large focal spot, and phantom images were simultaneously acquired using a micro focal spot as ground-truth. Peak-to-noise ratio (PSNR) and structural similarity (SSIM) were determined for quantitative analysis. The deblurring results were compared with the original data, and the image quality was subjectively evaluated and graded by two clinicians.

RESULTS

For quantitative analysis of phantom images, the average PSNR/SSIM based on the deep-learning approach (35.34/0.9566) was better than that of large focal spot images (30.64/0.9163). For subjective evaluation of 80 clinical patient images, image quality in all types of cerebrovascular diseases was also improved based on a deep-learning approach (p < 0.001).

CONCLUSIONS

Deep learning-based focal spot deblur algorithm can efficiently improve DSA image quality for better visualization of blood vessels and lesions in the image.

Artificial Intelligence and Deep Learning in Revolutionizing Brain Tumor Diagnosis and Treatment: A Narrative Review

The emergence of artificial intelligence (AI) in the medical field holds promise in improving medical management, particularly in personalized strategies for the diagnosis and treatment of brain tumors. However, integrating AI into clinical practice has proven to be a challenge. Deep learning (DL) is very convenient for extracting relevant information from large amounts of data that has increased in medical history and imaging records, which shortens diagnosis time, that would otherwise overwhelm manual methods. In addition, DL aids in automated tumor segmentation, classification, and diagnosis. DL models such as the Brain Tumor Classification Model and the Inception-Resnet V2, or hybrid techniques that enhance these functions and combine DL networks with support vector machine and k-nearest neighbors, identify tumor phenotypes and brain metastases, allowing real-time decision-making and enhancing preoperative planning. AI algorithms and DL development facilitate radiological diagnostics such as computed tomography, positron emission tomography scans, and magnetic resonance imaging (MRI) by integrating two-dimensional and three-dimensional MRI using DenseNet and 3D convolutional neural network architectures, which enable precise tumor delineation. DL offers benefits in neuro-interventional procedures, and the shift toward computer-assisted interventions acknowledges the need for more accurate and efficient image analysis methods. Further research is needed to realize the potential impact of DL in improving these outcomes.

Clinical Efficiency of an Artificial Intelligence-Based 3D-Angiography for Visualization of Cerebral Aneurysm: Comparison with the Conventional Method

PURPOSE

Artificial intelligence (AI)-based three-dimensional angiography (3D-A) was reported to demonstrate visualization of cerebral vasculature equivalent to that of three-dimensional digital subtraction angiography (3D-DSA). However, the applicability and efficacy of the AI-based 3D‑A algorithm have not yet been investigated for 3D-DSA micro imaging. In this study, we evaluated the usefulness of the AI-based 3D‑A in 3D-DSA micro imaging.

MATERIALS AND METHODS

The 3D-DSA micro datasets of 20 consecutive patients with cerebral aneurysm (CA) were reconstructed with 3D-DSA and 3D‑A. Three reviewers compared 3D-DSA and 3D‑A in terms of qualitative parameters (degrees of visualization of CA and the anterior choroidal artery [AChA]) and quantitative parameters (aneurysm diameter, neck diameter, parent vessel diameter, and visible length of AChA).

RESULTS

Qualitative evaluation of diagnostic potential revealed that visualization of CA and the proximal to middle parts of the AChA with 3D‑A was equal to that with conventional 3D-DSA; in contrast, visualization of the distal part of the AChA was lower with 3D‑A than with 3D-DSA. Further, regarding quantitative evaluation, the aneurysm diameter, neck diameter, and parent vessel diameter were comparable between 3D‑A and 3D-DSA; in contrast, the visible length of the AChA was lower with 3D‑A than with 3D-DSA.

CONCLUSIONS

The AI-based 3D‑A technique is feasible and evaluable visualization of cerebral vasculature with respect to quantitative and qualitative parameters in 3D-DSA micro imaging. However, the 3D‑A technique offers lower visualization of such as the distal portion of the AChA than 3D-DSA.

The Utility of Multimodal Imaging and Artificial Intelligence Algorithms for Overlying Two Volumes in the Decision Chain for the Treatment of Complex Pathologies in Interventional Neuroradiology—A Case Series Study

3D rotational angiography is now increasingly used in routine neuroendovascular procedures––in particular, for situations where the analysis of two overlayed sets of volume imaging proves useful for planning the treatment strategy or for confirming the optimal apposition of the intravascular devices used. The aim of this study is to identify and describe the decision algorithm for which the overlay function of 3D rotational angiography volumes, high-resolution contrast-enhanced flat panel detector CT adapted for intravascular devices (VasoCT/DynaCT), non-enhanced flat detector C-arm volume acquisition functionality integrated with the angiography equipment (XperCT/DynaCT), and isovolumetric MRI volumes were all used in treatments performed in a series of 29 patients. Two superposed 3DRA volumes were used in the treatment aneurysms located at the junction of two vascular territories and for arteriovenous malformations with compartments fed from different vascular territories. The superposition function of a preoperatively acquired 3DRA volume and a postoperatively acquired VasoCT volume provides accurate information about the apposition of neuroendovascular endoprostheses used in the treatment of aneurysms. The automatic overlay function generated by the 3D workstation is particularly useful, but in about 50% of cases it requires manual operator-dependent correction, requiring a certain level of experience. In our experience, multimodal imaging brings an important benefit, both in the treatment decision algorithm and in the assessment of neuroendovascular treatment efficacy.

Accuracy of Dose-Saving Artificial-Intelligence-Based 3D Angiography (3DA) for Grading of Intracranial Artery Stenoses: Preliminary Findings

BACKGROUND AND PURPOSE

Based on artificial intelligence (AI), 3D angiography (3DA) is a novel postprocessing algorithm for "DSA-like" 3D imaging of cerebral vasculature. Because 3DA requires neither mask runs nor digital subtraction as the current standard 3D-DSA does, it has the potential to cut the patient dose by 50%. The object was to evaluate 3DA's diagnostic value for visualization of intracranial artery stenoses (IAS) compared to 3D-DSA.

MATERIALS AND METHODS

3D-DSA datasets of IAS (n_IAS = 10) were postprocessed using conventional and prototype software (Siemens Healthineers AG, Erlangen, Germany). Matching reconstructions were assessed by two experienced neuroradiologists in consensus reading, considering image quality (IQ), vessel diameters (VD_1/2), vessel-geometry index (VGI = VD₁/VD₂), and specific qualitative/quantitative parameters of IAS (e.g., location, visual IAS grading [low-/medium-/high-grade] and intra-/poststenotic diameters [d_{intra-/poststenotic} in mm]). Using the NASCET criteria, the percentual degree of luminal restriction was calculated.

RESULTS

In total, 20 angiographic 3D volumes (n_3DA = 10; n_3D-DSA = 10) were successfully reconstructed with equivalent IQ. Assessment of the vessel geometry in 3DA datasets did not differ significantly from 3D-DSA (VD₁: r = 0.994, p = 0.0001; VD₂:r = 0.994, p = 0.0001; VGI: r = 0.899, p = 0.0001). Qualitative analysis of IAS location (3DA/3D-DSA:n_ICA/C4 = 1, n_ICA/C7 = 1, n_MCA/M1 = 4, n_VA/V4 = 2, n_BA = 2) and the visual IAS grading (3DA/3D-DSA:n_low-grade = 3, n_medium-grade = 5, n_high-grade = 2) revealed identical results for 3DA and 3D-DSA, respectively. Quantitative IAS assessment showed a strong correlation regarding intra-/poststenotic diameters (r_{dintrastenotic} = 0.995, p_{dintrastenotic} = 0.0001; r_{dpoststenotic} = 0.995, p_{dpoststenotic} = 0.0001) and the percentual degree of luminal restriction (r_{NASCET 3DA} = 0.981; p_{NASCET 3DA} = 0.0001).

CONCLUSIONS

The AI-based 3DA is a resilient algorithm for the visualization of IAS and shows comparable results to 3D-DSA. Hence, 3DA is a promising new method that allows a considerable patient-dose reduction, and its clinical implementation would be highly desirable.

Robotics and Artificial Intelligence in Endovascular Neurosurgery

The use of artificial intelligence (AI) and robotics in endovascular neurosurgery promises to transform neurovascular care. We present a review of the recently published neurosurgical literature on artificial intelligence and robotics in endovascular neurosurgery to provide insights into the current advances and applications of this technology.

The PubMed database was searched for "neurosurgery" OR "endovascular" OR "interventional" AND "robotics" OR "artificial intelligence" between January 2016 and August 2021. A total of 1296 articles were identified, and after applying the inclusion and exclusion criteria, 38 manuscripts were selected for review and analysis. These manuscripts were divided into four categories: 1) robotics and AI for the diagnosis of cerebrovascular pathology, 2) robotics and AI for the treatment of cerebrovascular pathology, 3) robotics and AI for training in neuroendovascular procedures, and 4) robotics and AI for clinical outcome optimization.

The 38 articles presented include 23 articles on AI-based diagnosis of cerebrovascular disease, 10 articles on AI-based treatment of cerebrovascular disease, two articles on AI-based training techniques for neuroendovascular procedures, and three articles reporting AI prediction models of clinical outcomes in vascular disorders of the brain. Innovation with robotics and AI focus on diagnostic efficiency, optimizing treatment and interventional procedures, improving physician procedural performance, and predicting clinical outcomes with the use of artificial intelligence and robotics. Experimental studies with robotic systems have demonstrated safety and efficacy in treating cerebrovascular disorders, and novel microcatheterization techniques may permit access to deeper brain regions. Other studies show that pre-procedural simulations increase overall physician performance. Artificial intelligence also shows superiority over existing statistical tools in predicting clinical outcomes.

The recent advances and current usage of robotics and AI in the endovascular neurosurgery field suggest that the collaboration between physicians and machines has a bright future for the improvement of patient care. The aim of this work is to equip the medical readership, in particular the neurosurgical specialty, with tools to better understand and apply findings from research on artificial intelligence and robotics in endovascular neurosurgery.

Mask-less Two-dimensional Digital Subtraction Angiography Generation Model for Abdominal Vasculature using Deep Learning

PURPOSE

To develop a deep learning model to generate synthetic, two-dimensional subtraction angiography images free of artifacts from native abdominal angiograms.

MATERIALS AND METHODS

In this retrospective study, two-dimensional digital subtraction angiograms (2D-DSA) and native angiograms were consecutively collected from July 2019 to March 2020. Images were divided into motion-free (training, validation, and motion-free test datasets) and containing motion artifacts (motion-artifact test dataset) sets. A total of 3185, 393, 383, and 345 images from 87 patients (mean age, 71 ± 10 years; 64 men, 23 women) were included in the training, validation, motion-free, and motion-artifacts test datasets, respectively. Native angiograms and 2D-DSA image pairs were used to train and validate an image-to-image translation model to generate synthetic deep learning-based subtraction angiography (DLSA) images. DLSA images were quantitatively evaluated by peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) using the motion-free dataset and were qualitatively evaluated by visual assessments by radiologists with a numerical rating scale using the motion-artifacts dataset.

RESULTS

The DLSA images showed mean PSNR (± standard deviation) of 43.05 ± 3.65 dB and mean SSIM of 0.98 ± 0.01, indicating high agreement with the original 2D-DSA images in the motion-free dataset. Qualitative visual evaluation by radiologists on the motion-artifacts dataset showed that DLSA images contained fewer motion artifacts than 2D-DSA. Additionally, DLSA images scored similarly to or higher than 2D-DSA images for vascular visualization and clinical usefulness.

CONCLUSION

The developed deep learning model could generate synthetic, motion-free subtraction images from abdominal angiograms with similar imaging characteristics to 2D-DSA images.

The Value of CT Angiography Based on Intelligent Segmentation Algorithm for Survival of Hemodialysis Patients

This study was to explore the application value for central venous stenosis and occlusion in hemodialysis patients under the CT angiography based on intelligent segmentation algorithm, so that patients can survive better. Spiral CT was used to examine upper limb swelling in 62 uremic hemodialysis patients at a speed of 3.8 mL/s. Nonionic iodine contrast agent was injected around the contralateral limb. The total dosage of 90-102 mL, it was scanned by intelligent trigger technology. The trigger scanning threshold was set. The monitoring point was located in the superior vena cava. CT with convolutional neural network intelligent segmentation algorithm was used to process image data. Finally, the quality of life and related biochemical levels of patients before and after hemodialysis were detected. Under the CT angiography of intelligent segmentation algorithm, 77 stenoses were found in 62 uremic patients, including 48 stenoses of the brachial vein and 17 stenoses of the superior vena cava. The correlation coefficient between CT angiography and digital subtraction angiography (DSA) imaging results of intelligent segmentation algorithm was 0.411. Segmentation effect of the algorithm in this study: automatic segmentation accuracy was greater than 79%. After hemodialysis treatment, the scores of physical fitness, pain, social function, and energy status of patients were significantly increased compared with those before treatment, and the levels of albumin, serum phosphorus, and parathyroid hormone were significantly decreased (P < 0.05). In summary, CT angiography with intelligent segmentation algorithm can obtain clear, intuitive, and complete vascular walking images, and better display subclavian vein, brachiocephalic vein, and superior vena cava. It can provide more valuable support for surgical intervention and has certain application value for better survival of hemodialysis patients.

Construction and imaging of a neurovascular unit model

In 2001, the concept of the neurovascular unit was introduced at the Stroke Progress Review Group meeting. The neurovascular unit is an important element of the health and disease status of blood vessels and nerves in the central nervous system. Since then, the neurovascular unit has attracted increasing interest from research teams, who have contributed greatly to the prevention, treatment, and prognosis of stroke and neurodegenerative diseases. However, additional research is needed to establish an efficient, low-cost, and low-energy in vitro model of the neurovascular unit, as well as enable noninvasive observation of neurovascular units in vivo and in vitro. In this review, we first summarize the composition of neurovascular units, then investigate the efficacy of different types of stem cells and cell culture methods in the construction of neurovascular unit models, and finally assess the progress of imaging methods used to observe neurovascular units in recent years and their positive role in the monitoring and investigation of the mechanisms of a variety of central nervous system diseases.

Low-Dose Three-Dimensional Rotational Angiography for Evaluating Intracranial Aneurysms: Analysis of Image Quality and Radiation Dose

Objective

This study aimed to evaluate the image quality and dose reduction of low-dose three-dimensional (3D) rotational angiography (RA) for evaluating intracranial aneurysms.

Materials and Methods

We retrospectively evaluated the clinical data and 3D RA datasets obtained from 146 prospectively registered patients (male:female, 46:100; median age, 58 years; range, 19–81 years). The subjective image quality of 79 examinations obtained from a conventional method and 67 examinations obtained from a low-dose (5-seconds and 0.10-µGy/frame) method was assessed by two neurointerventionists using a 3-point scale for four evaluation criteria. The total image quality score was then obtained as the average of the four scores. The image quality scores were compared between the two methods using a noninferiority statistical testing, with a margin of -0.2 (i.e., score of low-dose group – score of conventional group). For the evaluation of dose reduction, dose-area product (DAP) and air kerma (AK) were analyzed and compared between the two groups.

Results

The mean total image quality score ± standard deviation of the 3D RA was 2.97 ± 0.17 by reader 1 and 2.95 ± 0.20 by reader 2 for conventional group and 2.92 ± 0.30 and 2.95 ± 0.22, respectively, for low-dose group. The image quality of the 3D RA in the low-dose group was not inferior to that of the conventional group according to the total image quality score as well as individual scores for the four criteria in both readers. The mean DAP and AK per rotation were 5.87 Gy-cm² and 0.56 Gy, respectively, in the conventional group, and 1.32 Gy-cm² (p < 0.001) and 0.17 Gy (p < 0.001), respectively, in the low-dose group.

Conclusion

Low-dose 3D RA was not inferior in image quality and reduced the radiation dose by 70%–77% compared to the conventional 3D RA in evaluating intracranial aneurysms.

Artificial Intelligence-Based 3D Angiography for Visualization of Complex Cerebrovascular Pathologies

BACKGROUND AND PURPOSE

By means of artificial intelligence, 3D angiography is a novel postprocessing method for 3D imaging of cerebral vessels. Because 3D angiography does not require a mask run like the current standard 3D-DSA, it potentially offers a considerable reduction of the patient radiation dose. Our aim was an assessment of the diagnostic value of 3D angiography for visualization of cerebrovascular pathologies.

MATERIALS AND METHODS

3D-DSA data sets of cerebral aneurysms (n _CA = 10), AVMs (n _AVM = 10), and dural arteriovenous fistulas (dAVFs) (n _dAVF = 10) were reconstructed using both conventional and prototype software. Corresponding reconstructions have been analyzed by 2 neuroradiologists in a consensus reading in terms of image quality, injection vessel diameters (vessel diameter [VD] 1/2), vessel geometry index (VGI = VD1/VD2), and specific qualitative/quantitative parameters of AVMs (eg, location, nidus size, feeder, associated aneurysms, drainage, Spetzler-Martin score), dAVFs (eg, fistulous point, main feeder, diameter of the main feeder, drainage), and cerebral aneurysms (location, neck, size).

RESULTS

In total, 60 volumes have been successfully reconstructed with equivalent image quality. The specific qualitative/quantitative assessment of 3D angiography revealed nearly complete accordance with 3D-DSA in AVMs (eg, mean nidus size_{3D angiography/3D-DSA}= 19.9 [SD, 10.9]/20.2 [SD, 11.2] mm; r = 0.9, P = .001), dAVFs (eg, mean diameter of the main feeder_{3D angiography/3D-DSA}= 2.04 [SD, 0.65]/2.05 [SD, 0.63] mm; r = 0.9, P = .001), and cerebral aneurysms (eg, mean size_{3D angiography/3D-DSA}= 5.17 [SD, 3.4]/5.12 [SD, 3.3] mm; r = 0.9, P = .001). Assessment of the geometry of the injection vessel in 3D angiography data sets did not differ significantly from that of 3D-DSA (vessel geometry index_AVM: r = 0.84, P = .003; vessel geometry index_dAVF: r = 0.82, P = .003; vessel geometry index_CA: r = 0.84, P <.001).

CONCLUSIONS

In this study, the artificial intelligence-based 3D angiography was a reliable method for visualization of complex cerebrovascular pathologies and showed results comparable with those of 3D-DSA. Thus, 3D angiography is a promising postprocessing method that provides a significant reduction of the patient radiation dose.

Relaxation-Enhanced Angiography Without Contrast and Triggering (REACT) for Fast Imaging of Extracranial Arteries in Acute Ischemic Stroke at 3 T

Purpose

To evaluate a novel flow-independent 3D isotropic REACT sequence compared with CE-MRA for the imaging of extracranial arteries in acute ischemic stroke (AIS).

Methods

This was a retrospective study of 35 patients who underwent a stroke protocol at 3 T including REACT (fixed scan time: 2:46 min) and CE-MRA of the extracranial arteries. Three radiologists evaluated scans regarding vessel delineation, signal, and contrast and assessed overall image noise and artifacts using 5-point scales (5: excellent delineation/no artifacts). Apparent signal- and contrast-to-noise ratios (aSNR/aCNR) were measured for the common carotid artery (CCA), internal carotid artery (ICA, C1 segment), and vertebral artery (V2 segment). Two radiologists graded the degree of proximal ICA stenosis.

Results

Compared to REACT, CE-MRA showed better delineation for the CCA and ICA (C1 and C2 segments) (median 5, range 2–5 vs. 4, range 3–5; P < 0.05). For the ICA (C1 and C2 segments), REACT provided a higher signal (5, range 3–5; P < 0.05/4.5, range 3–5; P > 0.05 vs. 4, range 2–5) and contrast (5, range 3–5 vs. 4, range 2–5; P > 0.05) than CE-MRA. The remaining segments of the blood-supplying vessels showed equal medians. There was no significant difference regarding artifacts, whereas REACT provided significantly lower image noise (4, range 3–5 vs. 4 range 2–5; P < 0.05) with a higher aSNR (P < 0.05) and aCNR (P < 0.05) for all vessels combined. For clinically relevant (≥50%) ICA stenosis, REACT achieved a detection sensitivity of 93.75% and a specificity of 100%.

Conclusion

Given its fast acquisition, comparable image quality to CE-MRA and high sensitivity and specificity for the detection of ICA stenosis, REACT was proven to be a clinically applicable method to assess extracranial arteries in AIS.

Electronic supplementary material

The online version of this article (10.1007/s00062-020-00963-6) contains supplementary material, which is available to authorized users.