Counterfactual image generation enables controlled data augmentation, bias mitigation, and disease modeling. However, existing methods guided by external classifiers or regressors are limited to subject-level factors (e.g., age) and fail to produce localized structural changes, often resulting in global artifacts. Pixel-level guidance using segmentation masks has been explored, but requires user-defined counterfactual masks, which are tedious and impractical. Segmentor-guided Counterfactual Fine-Tuning (Seg-CFT) addressed this by using segmentation-derived measurements to supervise structure-specific variables, yet it remains restricted to global interventions. We propose Positional Seg-CFT, which subdivides each structure into regional segments and derives independent measurements per region, enabling spatially localized and anatomically coherent counterfactuals. Experiments on coronary CT angiography show that Pos-Seg-CFT generates realistic, region-specific modifications, providing finer spatial control for modeling disease progression.

BackgroundA patient-specific biomechanical model for quantifying the stress state within coronary atherosclerotic lesions for the noninvasive prediction of acute coronary syndrome (ACS) has not been established.MethodsSixty-nine patients with clearly documented ACS and available coronary computed tomographic angiography (cCTA) acquired between 1 month and 2 years before the development of ACS were included. In 63 culprit and 146 nonculprit lesions, lumen geometry, the presence of adverse plaque characteristics (APC), and hemodynamic parameters were evaluated using previously described methods. A novel biomechanical metric, denoted as the Cap Vulnerability Index (CVI), was calculated from cCTA data using computational solid and fluid mechanics and compared to anatomic, plaque and hemodynamics variables for the ability to discriminate culprit from nonculprit lesions.ResultsCulprit lesions had a greater diameter stenosis, lesion length, more adverse plaque and hemodynamic characteristics and a greater CVI. Among all parameters evaluated, CVI had the highest AUC (0.75), was the strongest independent predictor for ACS and was comparable to previously described statistical models combining anatomy, plaque and hemodynamic variables.ConclusionsNoninvasive biomechanical assessment of coronary plaques using computational solid and fluid mechanics may improve identification of culprit lesions for ACS and simplify the interpretation of risk factors for individual lesions.

Counterfactual image generation is a powerful tool for augmenting training data, de-biasing datasets, and modeling disease. Current approaches rely on external classifiers or regressors to increase the effectiveness of subject-level interventions (e.g., changing the patient's age). For structure-specific interventions (e.g., changing the area of the left lung in a chest radiograph), we show that this is insufficient, and can result in undesirable global effects across the image domain. Previous work used pixel-level label maps as guidance, requiring a user to provide hypothetical segmentations which are tedious and difficult to obtain. We propose Segmentor-guided Counterfactual Fine-Tuning (Seg-CFT), which preserves the simplicity of intervening on scalar-valued, structure-specific variables while producing locally coherent and effective counterfactuals. We demonstrate the capability of generating realistic chest radiographs, and we show promising results for modeling coronary artery disease. Code: https://github.com/biomedia-mira/seg-cft.

Coronary computed tomography angiography (CCTA) provides 3D information on obstructive coronary artery disease, but cannot fully visualize high-resolution features within the vessel wall. Intravascular imaging, in contrast, can spatially resolve atherosclerotic in cross sectional slices, but is limited in capturing 3D relationships between each slice. Co-registering CCTA and intravascular images enables a variety of clinical research applications but is time consuming and user-dependent. This is due to intravascular images suffering from non-rigid distortions arising from irregularities in the imaging catheter path. To address these issues, we present a morphology-based framework for the rigid and non-rigid matching of intravascular images to CCTA images. To do this, we find the optimal virtual catheter path that samples the coronary artery in CCTA image space to recapitulate the coronary artery morphology observed in the intravascular image. We validate our framework on a multi-center cohort of 40 patients using bifurcation landmarks as ground truth for longitudinal and rotational registration. Our registration approach significantly outperforms other approaches for bifurcation alignment. By providing a differentiable framework for multi-modal vascular co-registration, our framework reduces the manual effort required to conduct large-scale multi-modal clinical studies and enables the development of machine learning-based co-registration approaches.

Counterfactual image generation is a powerful tool for augmenting training data, de-biasing datasets, and modeling disease. Current approaches rely on external classifiers or regressors to increase the effectiveness of subject-level interventions (e.g., changing the patient's age). For structure-specific interventions (e.g., changing the area of the left lung in a chest radiograph), we show that this is insufficient, and can result in undesirable global effects across the image domain. Previous work used pixel-level label maps as guidance, requiring a user to provide hypothetical segmentations which are tedious and difficult to obtain. We propose Segmentor-guided Counterfactual Fine-Tuning (Seg-CFT), which preserves the simplicity of intervening on scalar-valued, structure-specific variables while producing locally coherent and effective counterfactuals. We demonstrate the capability of generating realistic chest radiographs, and we show promising results for modeling coronary artery disease. Code: https://github.com/biomedia-mira/seg-cft.

AIMS:Coronary computed tomography angiography provides non-invasive assessment of coronary stenosis severity and flow impairment. Automated artificial intelligence (AI) analysis may assist in precise quantification and characterization of coronary atherosclerosis, enabling patient-specific risk determination and management strategies. This multicentre international study compared an automated deep learning-based method for segmenting coronary atherosclerosis in coronary computed tomography angiography (CCTA) against the reference standard of intravascular ultrasound (IVUS). METHODS AND RESULTS:The study included clinically stable patients with known coronary artery disease from 15 centres in the USA and Japan. An AI-enabled plaque analysis was utilized to quantify and characterize total plaque (TPV), vessel, lumen, calcified plaque (CP), non-calcified plaque (NCP), and low-attenuation plaque (LAP) volumes derived from CCTA and compared with IVUS measurements in a blinded, core laboratory-adjudicated fashion. In 237 patients, 432 lesions were assessed; mean lesion length was 24.5 mm, and mean IVUS-TPV was 186.0 mm3. AI-enabled plaque analysis on CCTA showed strong correlation and high accuracy when compared with IVUS; correlation coefficient, slope, and Y intercept for TPV were 0.91, 0.99, and 1.87, respectively; for CP volume 0.91, 1.05, and 5.32, respectively; and for NCP volume 0.87, 0.98, and 15.24, respectively. Bland-Altman analysis demonstrated strong agreement with little bias for these measurements. CONCLUSION:AI-enabled CCTA quantification and characterization of atherosclerosis demonstrated strong agreement with IVUS reference standard measurements. This tool may prove effective for accurate evaluation of coronary atherosclerotic burden and cardiovascular risk assessment.

Background:Advanced coronary plaque analysis by cardiac computed tomography (CT) has recently emerged as a promising technique for better prognostic stratification. However, this evaluation application in clinical practice is still uncertain.Case summary:In the present case, we described the clinical picture of a 44-year-old tennis player with ectopic ventricular beats in which cardiac CT enabled the identification of a non-obstructive but high-risk plaque on proximal left anterior descendent artery. The application of artificial intelligence (AI)-enhanced software enabled to better stratify the patients' risk. The present case describes how early identification of non-obstructive but high-risk coronary plaque evaluated by cardiac CT using AI-enhanced software enabled accurate and personalized risk assessment.Discussion:The main clinical message of this case report is that advanced plaque analysis by cardiac CT, especially when performed with AI-based software, may provide important prognostic information leading to a personalized preventive approach. Moreover, AI-based software may contribute to promote a routine evaluation of these important data already included in traditional cardiac CT.

Virtual interventions enable the physics-based simulation of device deployment within coronary arteries. This framework allows for counterfactual reasoning by deploying the same device in different arterial anatomies. However, current methods to create such counterfactual arteries face a trade-off between controllability and realism. In this study, we investigate how Latent Diffusion Models (LDMs) can custom synthesize coronary anatomy for virtual intervention studies based on mid-level anatomic constraints such as topological validity, local morphological shape, and global skeletal structure. We also extend diffusion model guidance strategies to the context of morpho-skeletal conditioning and propose a novel guidance method for continuous attributes that adaptively updates the negative guiding condition throughout sampling. Our framework enables the generation and editing of coronary anatomy in a controllable manner, allowing device designers to derive mechanistic insights regarding anatomic variation and simulated device deployment. Our code is available at https://github.com/kkadry/Morphoskel-Diffusion .

Patient-specific models of blood flow in the coronary arteries have entered clinical practice worldwide to aid in the diagnosis and management of heart disease. This technology leverages modern AI-based image segmentation methods to extract the geometry of the coronary arteries from noninvasive computed tomography volumetric imaging, computational physiology methods to define boundary conditions and computational fluid dynamic techniques to compute coronary artery flow and pressure. The methods described herein have been used to compute the fractional flow reserve (FFR) in the coronary arteries, namely the ratio of pressure in a coronary artery to the reference aortic pressure under conditions of maximum hyperemia induced by the intravenous administration of a vasodilator to mimic increased coronary artery blood flow as occurs during physical activity. The fractional flow reserve derived from coronary computed tomography angiography (CCTA) using patient-specific models of coronary artery blood flow is referred to as FFRCT and has been validated against invasive measurements using pressure wires in several hundred patients and subsequently evaluated in clinical trials in several thousand patients. Patient-specific models of coronary artery blood flow can be used to predict outcomes of coronary artery revascularization procedures, to identify patients at risk of having a heart attack, and to model tissue perfusion of the microcirculation by coupling with emerging models of cardiac function. This paper provides a state-of-the-art review of computational methods for, and applications of, quantifying blood flow in patient-specific models of the coronary arteries.

Deep learning models for semantic segmentation are able to learn powerful representations for pixel-wise predictions, but are sensitive to noise at test time and do not guarantee a plausible topology. Image registration models on the other hand are able to warp known topologies to target images as a means of segmentation, but typically require large amounts of training data, and have not widely been benchmarked against pixel-wise segmentation models. We propose Atlas-ISTN, a framework that jointly learns segmentation and registration on 2D and 3D image data, and constructs a population-derived atlas in the process. Atlas-ISTN learns to segment multiple structures of interest and to register the constructed, topologically consistent atlas labelmap to an intermediate pixel-wise segmentation. Additionally, Atlas-ISTN allows for test time refinement of the model's parameters to optimize the alignment of the atlas labelmap to an intermediate pixel-wise segmentation. This process both mitigates for noise in the target image that can result in spurious pixel-wise predictions, as well as improves upon the one-pass prediction of the model. Benefits of the Atlas-ISTN framework are demonstrated qualitatively and quantitatively on 2D synthetic data and 3D cardiac computed tomography and brain magnetic resonance image data, out-performing both segmentation and registration baseline models. Atlas-ISTN also provides inter-subject correspondence of the structures of interest, enabling population-level shape and motion analysis.

OBJECTIVES:The goal of this study was to investigate the association of stenosis and plaque features with myocardial ischemia and their prognostic implications. BACKGROUND:Various anatomic, functional, and morphological attributes of coronary artery disease (CAD) have been independently explored to define ischemia and prognosis. METHODS:A total of 1,013 vessels with fractional flow reserve (FFR) measurement and available coronary computed tomography angiography were analyzed. Stenosis and plaque features of the target lesion and vessel were evaluated by an independent core laboratory. Relevant features associated with low FFR (≤0.80) were identified by using machine learning, and their predictability of 5-year risk of vessel-oriented composite outcome, including cardiac death, target vessel myocardial infarction, or target vessel revascularization, were evaluated. RESULTS:The mean percent diameter stenosis and invasive FFR were 48.5 ± 17.4% and 0.81 ± 0.14, respectively. Machine learning interrogation identified 6 clusters for low FFR, and the most relevant feature from each cluster was minimum lumen area, percent atheroma volume, fibrofatty and necrotic core volume, plaque volume, proximal left anterior descending coronary artery lesion, and remodeling index (in order of importance). These 6 features showed predictability for low FFR (area under the receiver-operating characteristic curve: 0.797). The risk of 5-year vessel-oriented composite outcome increased with every increment of the number of 6 relevant features, and it had incremental prognostic value over percent diameter stenosis and FFR (area under the receiver-operating characteristic curve: 0.706 vs. 0.611; p = 0.031). CONCLUSIONS:Six functionally relevant features, including minimum lumen area, percent atheroma volume, fibrofatty and necrotic core volume, plaque volume, proximal left anterior descending coronary artery lesion, and remodeling index, help define the presence of myocardial ischemia and provide better prognostication in patients with CAD. (CCTA-FFR Registry for Risk Prediction; NCT04037163).

In this paper, we introduce and compare different approaches for incorporating shape prior information into neural network-based image segmentation. Specifically, we introduce the concept of template transformer networks, where a shape template is deformed to match the underlying structure of interest through an end-to-end trained spatial transformer network. This has the advantage of explicitly enforcing shape priors, and this is free of discretization artifacts by providing a soft partial volume segmentation. We also introduce a simple yet effective way of incorporating priors in the state-of-the-art pixel-wise binary classification methods such as fully convolutional networks and U-net. Here, the template shape is given as an additional input channel, incorporating this information significantly reduces false positives. We report results on synthetic data and sub-voxel segmentation of coronary lumen structures in cardiac computed tomography showing the benefit of incorporating priors in neural network-based image segmentation.

In this paper we introduce and compare different approaches for incorporating shape prior information into neural network based image segmentation. Specifically, we introduce the concept of template transformer networks where a shape template is deformed to match the underlying structure of interest through an end-to-end trained spatial transformer network. This has the advantage of explicitly enforcing shape priors and is free of discretisation artefacts by providing a soft partial volume segmentation. We also introduce a simple yet effective way of incorporating priors in state-of-the-art pixel-wise binary classification methods such as fully convolutional networks and U-net. Here, the template shape is given as an additional input channel, incorporating this information significantly reduces false positives. We report results on sub-voxel segmentation of coronary lumen structures in cardiac computed tomography showing the benefit of incorporating priors in neural network based image segmentation.

Mammographic risk scoring has commonly been automated by extracting a set of handcrafted features from mammograms, and relating the responses directly or indirectly to breast cancer risk. We present a method that learns a feature hierarchy from unlabeled data. When the learned features are used as the input to a simple classifier, two different tasks can be addressed: i) breast density segmentation, and ii) scoring of mammographic texture. The proposed model learns features at multiple scales. To control the models capacity a novel sparsity regularizer is introduced that incorporates both lifetime and population sparsity. We evaluated our method on three different clinical datasets. Our state-of-the-art results show that the learned breast density scores have a very strong positive relationship with manual ones, and that the learned texture scores are predictive of breast cancer. The model is easy to apply and generalizes to many other segmentation and scoring problems.

Mammographic density is a well-established risk factor for breast cancer. We investigated the association between three different methods of measuring density or parenchymal pattern/texture on digitized film-based mammograms, and examined to what extent textural features independently and jointly with density can improve the ability to identify screening women at increased risk of breast cancer.

PURPOSE: The goal of this work is to develop a method to identify women at high risk for having breast cancer that is easily missed in regular mammography screening. Such a method will provide a rationale for selecting women for adjunctive screening. It goes beyond current risk assessment models that are not specifically adapted to reduce the number of interval cancers. METHOD AND MATERIALS: From the Dutch breast cancer screening program we collected 109 cancers that were screen negative and subsequently appeared as interval cancers, and 327 age matched healthy controls. To obtain mammograms without signs of cancerous tissue, we took the contralateral mammograms. We developed a novel machine learning based method called convolutional sparse autoencoder (CSAE) to characterize mammographic texture. The CSAE was trained and tested on raw mammograms to separate interval cancers from controls in a five-fold cross validation. To assess the independency of the texture scores of breast density, density was determined for each image using Volpara. RESULTS: The odds ratios for interval cancer were 1.59 (95% CI: 0.76-3.32), 2.07 (1.02-4.20), and 3.14 (1.60-6.17) for quartile 2, 3 and 4 respectively, relative to quartile 1. Correlation between the texture scores and breast density was 0.59 (0.52-0.64). Breast density adjusted odds ratios, as determined with logistic regression, were 1.49 (0.71-3.13), 1.58 (0.75-3.33), and 1.97 (0.91-4.27). CONCLUSIONS: The CSAE texture score is independently associated with the risk of having a breast cancer that is missed in regular mammography screening. As such it offers …

Mammographic scoring of density and texture are established methods to relate to the risk of breast cancer. We present a method that learns descriptive features from unlabeled mammograms and, using these learned features as the input to a simple classifier, address the following tasks: i) breast tissue segmentation ii) scoring of percentage mammographic density (PMD), and iii) scoring of mammographic texture (MT). Our results suggest that the learned PMD scores correlate well to manual ones, and that the learned MT scores are more related to future cancer risk than both manual and automatic PMD scores.

Segmentation of vertebral contours is an essential task in the design of imaging biomarkers for osteoporosis based on vertebra shape or texture. In this paper, we propose a novel automatic segmentation technique which can optionally be constrained by the user. The proposed technique solves the segmentation problem in a hierarchical manner. In the first phase, a coarse estimate of the overall spine alignment and the vertebra locations is computed using a sampling scheme. These samples are used to initialize a second phase of active shape model search, under a nonlinear model of vertebra appearance. The search is constrained by a conditional shape model, based on the variability of the coarse spine location estimates. In supplement, we describe an approach for manual initialization of the segmentation procedure as a simple set of constraints on the fully automatic technique. The technique is evaluated on a data base of 157 manually annotated lumbar radiographs, resulting in a final mean point-to-contour error of \(0.81~\pm ~0.98\) mm for automatic segmentation. The results outperform the previous work in automatic vertebra segmentation in terms of both segmentation accuracy and failure rate, offering a both automatic and semi-automatic approach in one unifying framework.

Segmentation of anatomical structures in medical images is often based on a voxel/pixel classification approach. Deep learning systems, such as convolutional neural networks (CNNs), can infer a hierarchical representation of images that fosters categorization. We propose a novel system for voxel classification integrating three 2D CNNs, which have a one-to-one association with the xy, yz and zx planes of 3D image, respectively. We applied our method to the segmentation of tibial cartilage in low field knee MRI scans and tested it on 114 unseen scans. Although our method uses only 2D features at a single scale, it performs better than a state-of-the-art method using 3D multi-scale features. In the latter approach, the features and the classifier have been carefully adapted to the problem at hand. That we were able to get better results by a deep learning architecture that autonomously learns the features from the images is the main insight of this study.

Poster: ECR 2013 / C-2041 / Fully automatic of film-based breast percentage density separate out postmenopausal hormone replacement treatment effects as well as expert’s estimation by: Petersen1, K. Chernoff2, G. Karemore3, M. Nielsen2; 1Frederiksberg/DK, 2Copenhagen/DK, 3 Rodovre/DK