Current State and Future Perspectives of Artificial Intelligence for Automated Coronary Angiography Imaging Analysis in Patients with Ischemic Heart Disease.

PURPOSE OF REVIEW: Artificial intelligence (AI) applications in (interventional) cardiology continue to emerge. This review summarizes the current state and future perspectives of AI for automated imaging analysis in invasive coronary angiography (ICA). RECENT FINDINGS: Recently, 12 studies on AI for automated imaging analysis In ICA have been published. In these studies, machine learning (ML) models have been developed for frame selection, segmentation, lesion assessment, and functional assessment of coronary flow. These ML models have been developed on monocenter datasets (in range 31-14,509 patients) and showed moderate to good performance. However, only three ML models were externally validated. Given the current pace of AI developments for the analysis of ICA, less-invasive, objective, and automated diagnosis of CAD can be expected in the near future. Further research on this technology in the catheterization laboratory may assist and improve treatment allocation, risk stratification, and cath lab logistics by integrating ICA analysis with other clinical characteristics.

Noise Reduction in CT Using Learned Wavelet-Frame Shrinkage Networks.

Encoding-decoding (ED) CNNs have demonstrated state-of-the-art performance for noise reduction over the past years. This has triggered the pursuit of better understanding the inner workings of such architectures, which has led to the theory of deep convolutional framelets (TDCF), revealing important links between signal processing and CNNs. Specifically, the TDCF demonstrates that ReLU CNNs induce low-rankness, since these models often do not satisfy the necessary redundancy to achieve perfect reconstruction (PR). In contrast, this paper explores CNNs that do meet the PR conditions. We demonstrate that in these type of CNNs soft shrinkage and PR can be assumed. Furthermore, based on our explorations we propose the learned wavelet-frame shrinkage network, or LWFSN and its residual counterpart, the rLWFSN. The ED path of the (r)LWFSN complies with the PR conditions, while the shrinkage stage is based on the linear expansion of thresholds proposed Blu and Luisier. In addition, the LWFSN has only a fraction of the training parameters (<1%) of conventional CNNs, very small inference times, low memory footprint, while still achieving performance close to state-of-the-art alternatives, such as the tight frame (TF) U-Net and FBPConvNet, in low-dose CT denoising.

Conditional Generative Adversarial Networks for low-dose CT image denoising aiming at preservation of critical image content.

X-ray Computed Tomography (CT) is an imaging modality where patients are exposed to potentially harmful ionizing radiation. To limit patient risk, reduced-dose protocols are desirable, which inherently lead to an increased noise level in the reconstructed CT scans. Consequently, noise reduction algorithms are indispensable in the reconstruction processing chain. In this paper, we propose to leverage a conditional Generative Adversarial Networks (cGAN) model, to translate CT images from low-to-routine dose. However, when aiming to produce realistic images, such generative models may alter critical image content. Therefore, we propose to employ a frequency-based separation of the input prior to applying the cGAN model, in order to limit the cGAN to high-frequency bands, while leaving low-frequency bands untouched. The results of the proposed method are compared to a state-of-the-art model within the cGAN model as well as in a single-network setting. The proposed method generates visually superior results compared to the single-network model and the cGAN model in terms of quality of texture and preservation of fine structural details. It also appeared that the PSNR, SSIM and TV metrics are less important than a careful visual evaluation of the results. The obtained results demonstrate the relevance of defining and separating the input image into desired and undesired content, rather than blindly denoising entire images. This study shows promising results for further investigation of generative models towards finding a reliable deep learning-based noise reduction algorithm for low-dose CT acquisition.

Non contrast-enhanced MRA versus ultrasound blood vessel assessment to determine the choice of hemodialysis vascular access.

PURPOSE: The aim of this work was to establish the relationship between traditional blood vessel mapping for vascular access (VA) creation by B-mode ultrasound (US) and novel non contrast-enhanced magnetic resonance angiography (NCE-MRA), and to study the potential influence of the diameter assessment technique on the choice of hemodialysis vascular access. METHODS: A total of 27 end-stage renal-disease patients were included. They received routine US and a NCE-MRA examination of the upper extremity. Diameters were measured manually on US and semi-automatically on NCE-MRA. These measurements were statistically compared for the arteries and veins and for each measurement location. Furthermore, sensitivity and specificity of both modalities to predict VA location was investigated by comparison with an experienced surgeon. This analysis gave insight into the potential influence of vessel mapping modality on decision-making. RESULTS: Comparison of NCE-MRA with US for the arteries and veins, demonstrated a bias of 9% (limits -33%-78%) and 38% (limits -36%-198%), respectively. Statistically significant differences between the modalities on the individual locations were mainly found for the venous locations. The sensitivity and specificity for US to predict VA location was 1.0 and 0.74, respectively, while for NCE-MRA this was 0.88 and 0.39, respectively. CONCLUSIONS: The results obtained indicate that extreme caution should be exercised when replacing one diameter measurement modality with the other. A further need exists to improve both vessel mapping protocols to obtain a geometric description of the upper extremity vasculature regardless of acquisition modality.

The benefit of non contrast-enhanced magnetic resonance angiography for predicting vascular access surgery outcome: a computer model perspective.

INTRODUCTION: Vascular access (VA) surgery, a prerequisite for hemodialysis treatment of end-stage renal-disease (ESRD) patients, is hampered by complication rates, which are frequently related to flow enhancement. To assist in VA surgery planning, a patient-specific computer model for postoperative flow enhancement was developed. The purpose of this study is to assess the benefit of non contrast-enhanced magnetic resonance angiography (NCE-MRA) data as patient-specific geometrical input for the model-based prediction of surgery outcome. METHODS: 25 ESRD patients were included in this study. All patients received a NCE-MRA examination of the upper extremity blood vessels in addition to routine ultrasound (US). Local arterial radii were assessed from NCE-MRA and converted to model input using a linear fit per artery. Venous radii were determined with US. The effect of radius measurement uncertainty on model predictions was accounted for by performing Monte-Carlo simulations. The resulting flow prediction interval of the computer model was compared with the postoperative flow obtained from US. Patients with no overlap between model-based prediction and postoperative measurement were further analyzed to determine whether an increase in geometrical detail improved computer model prediction. RESULTS: Overlap between postoperative flows and model-based predictions was obtained for 71% of patients. Detailed inspection of non-overlapping cases revealed that the geometrical details that could be assessed from NCE-MRA explained most of the differences, and moreover, upon addition of these details in the computer model the flow predictions improved. CONCLUSIONS: The results demonstrate clearly that NCE-MRA does provide valuable geometrical information for VA surgery planning. Therefore, it is recommended to use this modality, at least for patients at risk for local or global narrowing of the blood vessels as well as for patients for whom an US-based model prediction would not overlap with surgical choice, as the geometrical details are crucial for obtaining accurate flow predictions.

Accuracy and precision of vessel area assessment: manual versus automatic lumen delineation based on full-width at half-maximum.

PURPOSE: To evaluate the accuracy and precision of manual and automatic blood vessel diameter measurements, a quantitative comparison was conducted, using both phantom and clinical 3D magnetic resonance angiography (MRA) data. Since diameters are often manually measured, which likely is influenced by operator dependency, automatic lumen delineation, based on the full-width at half-maximum (FWHM), could improve these measurements. MATERIALS AND METHODS: Manual and automatic diameter assessments were compared, using MRA data from a vascular phantom (geometry obtained with µCT) and clinical MRA data. The diameters were manually assessed by 15 MRA experts, using both caliper and contour tools. To translate the experimental results to clinical practice, the precision obtained using phantom data was compared to the precision obtained with clinical data. RESULTS: A diameter error <10% was obtained with resolutions above 2, 3, and 5 pixels/diameter for the automatic FWHM, contour, and caliper methods, respectively. Using phantom data, precision of the manual methods was low (error >20%), even at high resolutions, while precision for the automatic method was high (error <3%) when using more than 2 pixels/diameter. A similar trend was found with clinical data. CONCLUSION: The results obtained clearly demonstrate improvement in the accuracy and precision of vessel diameter measurements with use of the automatic FWHM-based method.

Interactive virtual probing of 4D MRI blood-flow.

Better understanding of hemodynamics conceivably leads to improved diagnosis and prognosis of cardiovascular diseases. Therefore, an elaborate analysis of the blood-flow in heart and thoracic arteries is essential. Contemporary MRI techniques enable acquisition of quantitative time-resolved flow information, resulting in 4D velocity fields that capture the blood-flow behavior. Visual exploration of these fields provides comprehensive insight into the unsteady blood-flow behavior, and precedes a quantitative analysis of additional blood-flow parameters. The complete inspection requires accurate segmentation of anatomical structures, encompassing a time-consuming and hard-to-automate process, especially for malformed morphologies. We present a way to avoid the laborious segmentation process in case of qualitative inspection, by introducing an interactive virtual probe. This probe is positioned semi-automatically within the blood-flow field, and serves as a navigational object for visual exploration. The difficult task of determining position and orientation along the view-direction is automated by a fitting approach, aligning the probe with the orientations of the velocity field. The aligned probe provides an interactive seeding basis for various flow visualization approaches. We demonstrate illustration-inspired particles, integral lines and integral surfaces, conveying distinct characteristics of the unsteady blood-flow. Lastly, we present the results of an evaluation with domain experts, valuing the practical use of our probe and flow visualization techniques.

Volume measurement of intracranial aneurysms from 3D rotational angiography: improvement of accuracy by gradient edge detection.

Manual volume measurement of intracranial aneurysms from 3D rotational angiography varies on different threshold settings and, therefore, is operator-dependent. We developed and validated a method based on automatic gradient edge detection that is independent on threshold settings and provides an accurate and reproducible volume measurement. This method was compared with manual volume calculation in 13 aneurysm phantoms, and the results were significantly more accurate.

Relation between aneurysm volume, packing, and compaction in 145 cerebral aneurysms treated with coils.

PURPOSE: To assess the relation between aneurysm volume, packing, and compaction in cerebral aneurysms treated with coils. MATERIALS AND METHODS: The volumes of 145 aneurysms that were treated with coils were calculated with biplanar angiographic images and a custom-designed method. Partially thrombosed aneurysms were excluded. Packing was defined as the ratio between the volume of the inserted coils and the volume of the aneurysm and was calculated for all 145 aneurysms. Results at 6-month follow-up angiography were dichotomized into presence or absence of compaction. RESULTS: Aneurysm volume, packing, and compaction at 6-month follow-up were closely related. Large aneurysm volume was associated with low packing and frequent compaction. High packing prevents compaction. If the aneurysm volume was packed for 24% or more with coils, compaction did not occur in aneurysms with a volume of less than 600 mm(3). In small aneurysms with volumes of less than 200 mm(3), compaction did not occur when packing was above 20%. CONCLUSION: The common practice of inserting as many coils as possible in cerebral aneurysms is sensible in trying to avoid compaction. In aneurysms with packing of 24% or more, no compaction occurred at 6-month angiographic follow-up. In aneurysms with a volume of more than 600 mm(3), high packing could not be achieved, which resulted in compaction in the majority of aneurysms.