updates

Author: mstaring

_bibliography/papers.bib

@@ -13,8 +13,8 @@ @article{Chen:2025
   pages =    {},
   year =     {2025},
   pdf =      {2025_j_ER.pdf},
-  html =     {},
-  arxiv =    {},
+  html =     {https://doi.org/10.1007/s00330-025-12187-8},
+  arxiv =    {2511.20926},
   code =     {},
   abstract = {<b>Objectives:</b> To evaluate a deep learning (DL) model for reducing the agent dose of contrast-enhanced T1-weighted MRI (T1ce) of the cerebellopontine angle (CPA) cistern.<br><b>Materials and methods:</b> In this multi-center retrospective study, T1 and T1ce of vestibular schwannoma (VS) patients were used to simulate low-dose T1ce with varying reductions of contrast agent dose. DL models were trained to restore standard-dose T1ce from the low-dose simulation. The image quality and segmentation performance of the DL-restored T1ce were evaluated. A head and neck radiologist was asked to rate DL-restored images in multiple aspects, including image quality and diagnostic characterization.<br><b>Results:</b> 203 MRI studies from 72 VS patients (mean age, 58.51 &pm; 14.73, 39 men) were evaluated. As the input dose increased, the structural similarity index measure of the restored T1ce increased from 0.639 &pm; 0.113 to 0.993 &pm; 0.009, and the peak signal-to-noise ratio increased from 21.6 &pm; 3.73 dB to 41.4 &pm; 4.84 dB. At 10% input dose, using DL-restored T1ce for segmentation improved the Dice from 0.673 to 0.734, the 95% Hausdorff distance from 2.38 mm to 2.07 mm, and the average surface distance from 1.00 mm to 0.59 mm. Both DL-restored T1ce from 10% and 30% input doses showed excellent image quality (3.09 &pm; 0.811 and 3.23 &pm; 0.685), with the latter being considered more informative (3.81 &pm; 0.664).<br><b>Conclusion:</b> The DL model improved the image quality of low-dose MRI of the CPA cistern, which makes lesion detection and diagnostic characterization possible with 10% - 30% of the standard dose.<br><b>Key points</b><br><b>Question</b> Deep learning models that aid in the reduction of contrast agent dose are not extensively evaluated for MRI of the cerebellopontine angle cistern.<br><b>Finding</b> Deep learning models restored the low-dose MRI of the cerebellopontine angle cistern, yielding images sufficient for vestibular schwannoma diagnosis and management.<br><b>Clinical relevance statement</b> Deep learning models make it possible to reduce the use of gadolinium-based contrast agents for contrast-enhanced MRI of the cerebellopontine angle cistern.},
 }
@@ -88,7 +88,7 @@ @article{Du2025
   abstract = {<b>Background:</b> Pulmonary embolism (PE) is a life-threatening condition requiring prompt diagnosis and treatment. Visual assessment of computed tomographic pulmonary angiography (CTPA) is the first-choice diagnostic tool. New imaging biomarkers could provide additional prognostic information for improved risk stratification. We hypothesized in this exploratory study, that contrast enhancement patterns in the aorta may contain such information.<br><b>Methods:</b> CTPA scans of 93 acute PE patients were analyzed retrospectively. Firstly, the aorta was segmented automatically by TotalSegmentator and its centerline was extracted. Subsequently, lines were fitted on intensities within a region of interest perpendicularly to the aorta centerline, from which three parameters were extracted: mean intensity, proximal intensity and contrast gradient. After confounder analysis, logistic regression with forward selection evaluated the predictive value of these parameters for 12 adverse outcomes (six short-term and six long-term).<br><b>Results:</b> Lung volume, aorta dimension and contrast delay were considered as possible confounders but were not selected by forward selection. Logistic regression (n = 93) showed that a less steep contrast gradient (decreasing by 10 Hounsfield unit/%) was associated with a reduction in odds of the following short-term adverse outcomes: 48.1% for intensive care unit admission (odds ratio [OR] = 0.519, 95% confidence interval [CI]: 0.306-0.804), 29.3% for oxygen therapy >24 hours (OR = 0.707, 95% CI: 0.496-0.976), 60.6% for reperfusion therapy (OR = 0.394, 95% CI: 0.178-0.682), 57.5% for vasopressor therapy (OR = 0.425, 95% CI: 0.194-0.741), and 50.2% for PE-related death (OR = 0.498, 95% CI: 0.246-0.915). No significant associations were found with long-term outcomes.<br><b>Conclusions:</b> The aorta contrast gradient, automatically quantified from CTPA, is a relevant adjunctive predictor for short-term outcomes in PE patients. Long-term outcomes, however, could not be predicted by aorta measurement. This pilot study provides initial insights into predictive value of aorta enhancement, stimulating further exploration with external data.},
 }
 
-@article{VanDerValk2025,
+@article{VanDerLoo2025,
   abbr =     {},
   bibtex_show = {true},
   title =   {Large Language Models for Structured Cardiovascular Data Extraction: A Foundation for Scalable Research and Clinical Applications},
@@ -99,7 +99,7 @@ @article{VanDerValk2025
   month =   {},
   year =    {2025},
   pdf =     {2025_j_EHJ-DH.pdf},
-  html =    {},
+  html =    {https://doi.org/10.1093/ehjdh/ztaf127},
   arxiv =   {},
   code =    {},
   abstract = {<b>Background:</b> Automated extraction of information from cardiac reports would benefit both clinical reporting and research. Large language models (LLMs) hold promise for such automation, but their clinical performance and practical implementation across various computational environments remain unclear.<br><b>Objectives:</b> To evaluate the feasibility and performance of LLM-based classification of echocardiogram and invasive coronary angiography (ICA) reports, using real-world clinical data across local, high-performance computing and cloud-based platforms.<br><b>Methods:</b> The angiography and echocardiography reports of 1000 patients, admitted with acute coronary syndrome, were labeled for multiple key diagnostic elements, including left ventricular function (LVF), culprit vessel and acute occlusions. Report classification models were developed using LLMs via i. prompt-based and ii. fine-tuning approaches. Performance was assessed across different model types and compute infrastructures, with attention to class imbalance, ambiguous label annotations and implementation costs.<br><b>Results:</b> LLMs demonstrated strong performance in extracting structured diagnostic information from cardiac reports. Cloud-based models (such as GPT-4o) achieved the highest accuracy (0.87 for culprit vessel and 1.0 for LVF) and generalizability, but also smaller models run on a local high performance cluster (HPC) achieved reasonable accuracy, especially for less complex tasks (0.634 for culprit vessel and 0.984 for LVF). Classification was feasible with minimal preprocessing, enabling potential integration into electronic health record systems or research pipelines. Class imbalance, reflective of real-world prevalence, had a greater impact on fine-tuning approaches.<br><b>Conclusions:</b> LLMs can reliably classify structured cardiology reports across diverse compute infrastructures. Their accuracy and adaptability support their use in clinical and research settings, particularly for scalable report structuring and dataset generation.},

_bibliography/papers_abstracts.bib

@@ -63,7 +63,7 @@ @inproceedings{Jabarimani:2025a
   month =     {May},
   year =      {2025},
   pdf =       {},
-  html =      {},
+  html =      {https://doi.org/10.58530/2025/1652},
   arxiv =     {},
   code =      {},
   abstract =  {Mid field MR scanners (0.1T-1T) are gaining increasing attention in the last few years (Lavrova et al. (2024), (Campbell Washburn et al., 2019) due to increased safety and lower manufacturing and maintenance cost and consequently improving the accessibility of MRI for clinical purposes (Arnold et al., 2023). The reduced field strength leads to a reduced signal and change in T1 relaxation times and therefore different contrast and signal to noise as 1.5T and 3T systems.<br>MRI is often used for diagnosing and monitoring brain tumors, lesions, and disorders such as neurodegenerative diseases. Especially for the last afflictions gray matter (GM) and white matter (WM) volume are interesting markers due to the atrophy associated with these diseases. Automatic segmentation models including Adaptive Maximum A Posteriori (MAP) segmentation (Rajapakse et al., 1997) and partial volume estimation (PVE) (Tohka et al., 2004) are well established models that are used by applications such as CAT12 to estimate GM and WM volumes.<br>In this work we study the quality of T1 weighted based brain segmentations at 0.6T in the context of brain volume measurements in healthy volunteers. We compare segmentations from 0.6T and 1.5T and study how noise suppression by deep learning based reconstruction as provided by the vendor might affect these.},
@@ -78,7 +78,7 @@ @inproceedings{Jabarimani:2025b
   month =     {May},
   year =      {2025},
   pdf =       {},
-  html =      {},
+  html =      {https://doi.org/10.58530/2025/2766},
   arxiv =     {},
   code =      {},
   abstract =  {Fluid-Attenuated Inversion Recovery (FLAIR) images are an important part of clinical brain protocols, especially due to the excellent contrast for diagnosing lesions, edema, etc. (Campbell-Washburn et al., 2019). Mid field MR scanners (0.1T-1T) provide a more accessible and affordable option for clinical use compared to high-field scanners (Arnold et al., 2023). However, lower field strength often produces lower-quality FLAIR images due to longer T1 relaxation time and lower signal to noise ratio (Lavrova et al., 2024). Moreover, from our initial tests with a 0.6T MRI-scanner, we noticed that the relative drop in quality of these images compared to 1.5T was much higher for FLAIR than for example T2W images, which could limit their diagnostic value.<br>This project aims to address this issue by applying a content/style-based plug-and-play reconstruction framework PnP-MUNIT (Rao et al., 2024) to guide the reconstruction of FLAIR images using information from T2-weighted scans, with the goal of improving image quality and reducing the scan time. In this work we assess this concept using data from a 3T scanner and adapt it for 0.6T data, applying the content/style model to the lower field strength in a zero-shot manner without any fine-tuning.},
@@ -93,7 +93,7 @@ @inproceedings{Gonzalez:2025
   month =     {May},
   year =      {2025},
   pdf =       {},
-  html =      {},
+  html =      {https://doi.org/10.58530/2025/1299},
   arxiv =     {},
   code =      {},
   abstract =  {The use of Imageless MR sequences, combined with deep-learning methods, could offer a rapid, cost-effective screening technique suitable for large population-wise deployment. We showcase how this framework yields accurate detection and lesion size estimation using an MS lesions case study.},
@@ -108,7 +108,7 @@ @inproceedings{Ilicak:2025
   month =     {May},
   year =      {2025},
   pdf =       {},
-  html =      {},
+  html =      {https://doi.org/10.58530/2025/1871},
   arxiv =     {},
   code =      {},
   abstract =  {We investigate a prototype 0.6T MRI system for free-breathing functional lung imaging. Our findings demonstrate improved image quality compared to 1.5T, with improved tissue-background contrast and homogeneity the functional maps, underscoring the system's robustness and potential for non-invasive pulmonary imaging.},
@@ -123,7 +123,7 @@ @inproceedings{Rao:2024
   month =     {May},
   year =      {2024},
   pdf =       {2024_a_ISMRM.pdf},
-  html =      {},
+  html =      {https://doi.org/10.58530/2024/0808},
   arxiv =     {},
   code =      {},
   abstract =  {<b>Motivation: </b>Scans within an MR exam share redundant information due to the same underlying structures. One contrast can hence be used to guide the reconstruction of another, thereby requiring less measurements.<br><b>Goals: </b>Multimodal guided reconstruction to reduce scanning times.<br><b>Approach: </b>Our method exploits AI-based content/style decomposition in an iterative reconstruction algorithm. We explored this concept via numerical simulation and subsequently validated it on in vivo data.<br><b>Results: </b>Compared to a conventional compressed sensing baseline, our method showed consistent improvement in simulations and produced sharper reconstructions from undersampled in vivo data. By enforcing data consistency, it was also more reliable than blind image translation.<br><b>Impact: </b>In the clinic, this can potentially enable a reduced MR exam time for a given image quality or improve image quality given a scan time budget. The former can reduce strain on the patient, whereas the latter can improve diagnosis.},

_bibliography/papers_conf.bib

@@ -22,6 +22,25 @@ @inproceedings{Zheng:2025b
   abstract =  {Encoder-Decoder architectures are widely used in deep learning-based Deformable Image Registration (DIR), where the encoder extracts multi-scale features and the decoder predicts deformation fields by recovering spatial locations. However, current methods lack specialized extraction of features (that are useful for registration) and predict deformation jointly and homogeneously in all three directions. In this paper, we propose a novel expert-guided DIR network with Mixture of Experts (MoE) mechanism applied in both encoder and decoder, named SHMoAReg. Specifically, we incorporate Mixture of Attention heads (MoA) into encoder layers, while Spatial Heterogeneous Mixture of Experts (SHMoE) into the decoder layers. The MoA enhances the specialization of feature extraction by dynamically selecting the optimal combination of attention heads for each image token. Meanwhile, the SHMoE predicts deformation fields heterogeneously in three directions for each voxel using experts with varying kernel sizes. Extensive experiments conducted on two publicly available datasets show consistent improvements over various methods, with a notable increase from 60.58% to 65.58% in Dice score for the abdominal CT dataset. Furthermore, SHMoAReg enhances model interpretability by differentiating experts' utilities across/within different resolution layers. To the best of our knowledge, we are the first to introduce MoE mechanism into DIR tasks. The code will be released soon.},
 }
 
+@inproceedings{Imre:2025,
+  abbr =    {},
+  bibtex_show = {true},
+  author =    {Imre, Baris and Rao, Chinmay and Salehi, Aram and Staring, Marius and Ilicak, Efe},
+  title =     {Bridged Denoising Diffusion in Ultra-Low Field MRI Enhancement Challenge},
+  booktitle = {MICCAI},
+  address =   {Daejeon, South Korea},
+  series =    {Lecture Notes in Computer Science},
+  volume =    {},
+  pages =     {},
+  month =     {September},
+  year =      {2025},
+  pdf =       {2025_c_MICCAIc.pdf},
+  html =      {},
+  arxiv =     {},
+  code =      {},
+  abstract =  {Low-field MRI offers a cost-effective and portable alternative to conventional high-field systems, but its clinical use is limited by reduced image quality. Recent advancements in deep learning have aimed to enhance low-field scans, including approaches based on GANs, stochastic quality transfer, and denoising diffusion models. In this paper, we introduce a bridged denoising diffusion model that explicitly aligns the latent spaces of low- and high-field MR images through a learned bridging model. At a predefined timestep in the diffusion process, a bridge network translates the noisy low-field representation into the high-field domain, enabling the downstream diffusion model, trained solely on high-field data, to generate high-field-like images. The proposed method is trained on paired 3T and 0.64 mT scans across multiple contrasts, based on the "Enhancing Ultra-Low-Field MRI with Paired High-Field MRI Comparisons for Brain Imaging" challenge. On the hidden validation set, our method achieved an SSIM of 0.779 and a PSNR of 22.85 dB, outperforming alternative configurations in ablation experiments. We show that our model can significantly enhance low-field images.},
+}
+
 @inproceedings{Lyu:2025b,
   abbr =    {},
   bibtex_show = {true},