Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention最新文献

Since its introduction, UNet has been leading a variety of medical image segmentation tasks. Although numerous follow-up studies have also been dedicated to improving the performance of standard UNet, few have conducted in-depth analyses of the underlying interest pattern of UNet in medical image segmentation. In this paper, we explore the patterns learned in a UNet and observe two important factors that potentially affect its performance: (i) irrelative feature learned caused by asymmetric supervision; (ii) feature redundancy in the feature map. To this end, we propose to balance the supervision between encoder and decoder and reduce the redundant information in the UNet. Specifically, we use the feature map that contains the most semantic information (i.e., the last layer of the decoder) to provide additional supervision to other blocks to provide additional supervision and reduce feature redundancy by leveraging feature distillation. The proposed method can be easily integrated into existing UNet architecture in a plug-and-play fashion with negligible computational cost. The experimental results suggest that the proposed method consistently improves the performance of standard UNets on four medical image segmentation datasets. The code is available at https://github.com/ChongQingNoSubway/SelfReg-UNet.

Asymptomatic neurocognitive impairment (ANI) is a predominant form of cognitive impairment among individuals infected with human immunodeficiency virus (HIV). The current diagnostic criteria for ANI primarily rely on subjective clinical assessments, possibly leading to different interpretations among clinicians. Some recent studies leverage structural or functional MRI containing objective biomarkers for ANI analysis, offering clinicians companion diagnostic tools. However, they mainly utilize a single imaging modality, neglecting complementary information provided by structural and functional MRI. To this end, we propose an attention-enhanced structural and functional MRI fusion (ASFF) framework for HIV-associated ANI analysis. Specifically, the ASFF first extracts data-driven and human-engineered features from structural MRI, and also captures functional MRI features via a graph isomorphism network and Transformer. A mutual cross-attention fusion module is then designed to model the underlying relationship between structural and functional MRI. Additionally, a semantic inter-modality constraint is introduced to encourage consistency of multimodal features, facilitating effective feature fusion. Experimental results on 137 subjects from an HIV-associated ANI dataset with T1-weighted MRI and resting-state functional MRI show the effectiveness of our ASFF in ANI identification. Furthermore, our method can identify both modality-shared and modality-specific brain regions, which may advance our understanding of the structural and functional pathology underlying ANI.

This work investigates the use of configuration state imaging together with deep neural networks to develop quantitative MRI techniques for deployment in an interventional setting. A physics modeling technique for inhomogeneous fields and heterogeneous tissues is presented and used to evaluate the theoretical capability of neural networks to estimate parameter maps from configuration state signal data. All tested normalization strategies achieved similar performance in estimating $T_2$ and $T_2^{*}$ . Varying network architecture and data normalization had substantial impacts on estimated flip angle and $T_1$ , highlighting their importance in developing neural networks to solve these inverse problems. The developed signal modeling technique provides an environment that will enable the development and evaluation of physics-informed machine learning techniques for MR parameter mapping and facilitate the development of quantitative MRI techniques to inform clinical decisions during MR-guided treatments.

Higher-order properties of functional magnetic resonance imaging (fMRI) induced connectivity have been shown to unravel many exclusive topological and dynamical insights beyond pairwise interactions. Nonetheless, whether these fMRI-induced higher-order properties play a role in disentangling other neuroimaging modalities' insights remains largely unexplored and poorly understood. In this work, by analyzing fMRI data from the Human Connectome Project Young Adult dataset using persistent homology, we discovered that the volume-optimal persistence homological scaffolds of fMRI-based functional connectomes exhibited conservative topological reconfigurations from the resting state to attentional task-positive state. Specifically, while reflecting the extent to which each cortical region contributed to functional cycles following different cognitive demands, these reconfigurations were constrained such that the spatial distribution of cavities in the connectome is relatively conserved. Most importantly, such level of contributions covaried with powers of aperiodic activities mostly within the theta-alpha (4-12 Hz) band measured by magnetoencephalography (MEG). This comprehensive result suggests that fMRI-induced hemodynamics and MEG theta-alpha aperiodic activities are governed by the same functional constraints specific to each cortical morpho-structure. Methodologically, our work paves the way toward an innovative computing paradigm in multimodal neuroimaging topological learning. The code for our analyses is provided in https://github.com/ngcaonghi/scaffold_noise.

Deep learning models generating structural brain MRIs have the potential to significantly accelerate discovery of neuroscience studies. However, their use has been limited in part by the way their quality is evaluated. Most evaluations of generative models focus on metrics originally designed for natural images (such as structural similarity index and Fréchet inception distance). As we show in a comparison of 6 state-of-the-art generative models trained and tested on over 3000 MRIs, these metrics are sensitive to the experimental setup and inadequately assess how well brain MRIs capture macrostructural properties of brain regions (a.k.a., anatomical plausibility). This shortcoming of the metrics results in inconclusive findings even when qualitative differences between the outputs of models are evident. We therefore propose a framework for evaluating models generating brain MRIs, which requires uniform processing of the real MRIs, standardizing the implementation of the models, and automatically segmenting the MRIs generated by the models. The segmentations are used for quantifying the plausibility of anatomy displayed in the MRIs. To ensure meaningful quantification, it is crucial that the segmentations are highly reliable. Our framework rigorously checks this reliability, a step often overlooked by prior work. Only 3 of the 6 generative models produced MRIs, of which at least 95% had highly reliable segmentations. More importantly, the assessment of each model by our framework is in line with qualitative assessments, reinforcing the validity of our approach. The code of this framework is available via https://github.com/jiaqiw01/MRIAnatEval.git.

In resource-limited settings, portable ultra-low-field (uLF, i.e., 0.064T) magnetic resonance imaging (MRI) systems expand accessibility of radiological scanning, particularly for low-income areas as well as underserved populations like neonates and infants. However, compared to high-field (HF, e.g., ≥ 1.5T) systems, inferior image quality in uLF scanning poses challenges for research and clinical use. To address this, we introduce Super-Field Network (SFNet), a custom swinUNETRv2 with generative adversarial network components that uses uLF MRIs to generate super-field (SF) images comparable to HF MRIs. We acquired a cohort of infant data (n=30, aged 0-2 years) with paired uLF-HF MRI data from a resource-limited setting with an underrepresented population in research. To enhance the small dataset, we present a novel use of latent diffusion to create dual-channel (uLF-HF) paired MRIs. We compare SFNet with state-of-the-art synthesis methods by HF-SF image similarity perceptual scores and by automated HF and SF segmentations of white matter (WM), gray matter (GM), and cerebrospinal fluid (CSF). The best performance was achieved by SFNet trained on the latent diffusion enhanced dataset yielding state-of-the-art results in Fréchet inception distance at 9.08 ± 1.21, perceptual similarity at 0.11 ± 0.01, and PSNR at 22.64 ± 1.31. True HF and SF segmentations had a strong overlap with Dice similarity coefficients of 0.71 ± 0.1, 0.79 ± 0.2, and 0.73 ± 0.08 for WM, GM, and CSF, respectively, in the developing infant brain with incomplete myelination, and displayed 166%, 107%, and 106% improvement over respective uLF-based segmentation metrics. SF MRI supports health equity by enhancing the clinical use of uLF imaging systems and improving the diagnostic capabilities of low-cost portable MRI systems in resource-limited settings and for underserved populations. Our code is made openly available at https://github.com/AustinTapp/SFnet.

Intraoperative ultrasound (iUS) imaging has the potential to improve surgical outcomes in brain surgery. However, its interpretation is challenging, even for expert neurosurgeons. In this work, we designed the first patient-specific framework that performs brain tumor segmentation in trackerless iUS. To disambiguate ultrasound imaging and adapt to the neurosurgeon's surgical objective, a patient-specific real-time network is trained using synthetic ultrasound data generated by simulating virtual iUS sweep acquisitions in pre-operative MR data. Extensive experiments performed in real ultrasound data demonstrate the effectiveness of the proposed approach, allowing for adapting to the surgeon's definition of surgical targets and outperforming non-patient-specific models, neurosurgeon experts, and high-end tracking systems. Our code is available at: https://github.com/ReubenDo/MHVAE-Seg.

Individualized brain parcellations derived from functional MRI (fMRI) are essential for discerning unique functional patterns of individuals, facilitating personalized diagnoses and treatments. Unfortunately, as fMRI signals are inherently noisy, establishing reliable individualized parcellations typically necessitates long-duration fMRI scan (> 25 min), posing a major challenge and resulting in the exclusion of numerous short-duration fMRI scans from individualized studies. To address this issue, we develop a novel Consecutive-Contrastive Spherical U-net (CC-SUnet) to enable the prediction of reliable individualized brain parcellation using short-duration fMRI data, greatly expanding its practical applicability. Specifically, 1) the widely used functional diffusion map (DM), obtained from functional connectivity, is carefully selected as the predictive feature, for its advantage in tracing the transitions between regions while reducing noise. To ensure a robust depiction of brain network, we propose a dual-task model to predict DM and cortical parcellation simultaneously, fully utilizing their reciprocal relationship. 2) By constructing a stepwise dataset to capture the gradual changes of DM over increasing scan durations, a consecutive prediction framework is designed to realize the prediction from short-to-long gradually. 3) A stepwise-denoising-prediction module is further proposed. The noise representations are separated and replaced by the latent representations of a group-level diffusion map, realizing informative guidance and denoising concurrently. 4) Additionally, an N-pair contrastive loss is introduced to strengthen the discriminability of the individualized parcellations. Extensive experimental results demonstrated the superiority of our proposed CC-SUnet in enhancing the reliability of the individualized parcellation with short-duration fMRI data, thereby significantly boosting their utility in individualized studies.

Graph neural networks (GNNs) are proficient machine learning models in handling irregularly structured data. Nevertheless, their generic formulation falls short when applied to the analysis of brain connectomes in Alzheimer's Disease (AD), necessitating the incorporation of domain-specific knowledge to achieve optimal model performance. The integration of AD-related expertise into GNNs presents a significant challenge. Current methodologies reliant on manual design often demand substantial expertise from external domain specialists to guide the development of novel models, thereby consuming considerable time and resources. To mitigate the need for manual curation, this paper introduces a novel self-guided knowledge-infused multimodal GNN to autonomously integrate domain knowledge into the model development process. We propose to conceptualize existing domain knowledge as natural language, and devise a specialized multimodal GNN framework tailored to leverage this uncurated knowledge to direct the learning of the GNN submodule, thereby enhancing its efficacy and improving prediction interpretability. To assess the effectiveness of our framework, we compile a comprehensive literature dataset comprising recent peer-reviewed publications on AD. By integrating this literature dataset with several real-world AD datasets, our experimental results illustrate the effectiveness of the proposed method in extracting curated knowledge and offering explanations on graphs for domain-specific applications. Furthermore, our approach successfully utilizes the extracted information to enhance the performance of the GNN.

Low-dose computed tomography (LDCT) reduces the risks of radiation exposure but introduces noise and artifacts into CT images. The Feature Pyramid Network (FPN) is a conventional method for extracting multi-scale feature maps from input images. While upper layers in FPN enhance semantic value, details become generalized with reduced spatial resolution at each layer. In this work, we propose a Gradient Guided Co-Retention Feature Pyramid Network (G2CR-FPN) to address the connection between spatial resolution and semantic value beyond feature maps extracted from LDCT images. The network is structured with three essential paths: the bottom-up path utilizes the FPN structure to generate the hierarchical feature maps, representing multi-scale spatial resolutions and semantic values. Meanwhile, the lateral path serves as a skip connection between feature maps with the same spatial resolution, while also functioning feature maps as directional gradients. This path incorporates a gradient approximation, deriving edge-like enhanced feature maps in horizontal and vertical directions. The top-down path incorporates a proposed co-retention block that learns the high-level semantic value embedded in the preceding map of the path. This learning process is guided by the directional gradient approximation of the high-resolution feature map from the bottom-up path. Experimental results on the clinical CT images demonstrated the promising performance of the model. Our code is available at: https://github.com/liz109/G2CR-FPN.