Pub Date : 2024-12-04DOI: 10.1016/j.mri.2024.110294
Frederick C Damen, Changliang Su, Jay Tsuruda, Thomas Anderson, Tibor Valyi-Nagy, Weiguo Li, Mehran Shaghaghi, Rifeng Jiang, Chuanmiao Xie, Kejia Cai
Background: In conjunction with an epidemiologically determined treatment window, current radiological acute ischemic stroke practice discerns two lesion (stage) types: core (dead tissue, identified by diffusion-weighted imaging (DWI)) and penumbra (tissue region receiving just enough blood flow to be potentially salvageable, identified by the perfusion diffusion mismatch). However, advancements in preclinical and clinical studies have indicated that this approach may be too rigid, warranting a more fine-grained patient-tailored approach. This study aimed to demonstrate the ability to noninvasively provide insights into the current in vivo stroke lesion cascade.
Methods: To elucidate a finer-grained depiction of the acute focal ischemic stroke cascade in vivo, we retrospectively applied our multimodal apparent diffusion (MAD) method to multi-b-value DWI, up to a b-value of 10,000 s/mm2 in 34 patients with acute focal ischemic stroke. Fuzzy C Means was used to cluster the MAD parameters.
Results: We discerned 18 clusters consistent with normal appearing tissue (NAT) types and 14 potential ischemic lesion (stage) types, providing insights into the variability and aggressiveness of lesion progression and current anomalous stroke-related imaging features. Of the 529 ischemic stroke lesion instances previously identified by two radiologists, 493 (92 %) were autonomously identified; 460 (87 %) were identified as efficaciously or better than the radiologists.
Conclusions: The data analyzed included a small number of clinical patients without follow-up or contemporaneous histology; therefor, the findings and theorizing should be treated as conjecture. Nevertheless, each identified NAT and lesion type is consistent with the known underpinnings of physiological tissues and pathological ischemic stroke lesion (stage) types. Several findings should be considered in current clinical imaging: WM fluid accumulation, BBB compromise conundrum, b1000 identified core may not be dead tissue, and a practical reason for DWI (pseudo) normalization.
{"title":"The fuzzy MAD stroke conjecture, using Fuzzy C Means to classify multimodal apparent diffusion for ischemic stroke lesion stratification.","authors":"Frederick C Damen, Changliang Su, Jay Tsuruda, Thomas Anderson, Tibor Valyi-Nagy, Weiguo Li, Mehran Shaghaghi, Rifeng Jiang, Chuanmiao Xie, Kejia Cai","doi":"10.1016/j.mri.2024.110294","DOIUrl":"10.1016/j.mri.2024.110294","url":null,"abstract":"<p><strong>Background: </strong>In conjunction with an epidemiologically determined treatment window, current radiological acute ischemic stroke practice discerns two lesion (stage) types: core (dead tissue, identified by diffusion-weighted imaging (DWI)) and penumbra (tissue region receiving just enough blood flow to be potentially salvageable, identified by the perfusion diffusion mismatch). However, advancements in preclinical and clinical studies have indicated that this approach may be too rigid, warranting a more fine-grained patient-tailored approach. This study aimed to demonstrate the ability to noninvasively provide insights into the current in vivo stroke lesion cascade.</p><p><strong>Methods: </strong>To elucidate a finer-grained depiction of the acute focal ischemic stroke cascade in vivo, we retrospectively applied our multimodal apparent diffusion (MAD) method to multi-b-value DWI, up to a b-value of 10,000 s/mm<sup>2</sup> in 34 patients with acute focal ischemic stroke. Fuzzy C Means was used to cluster the MAD parameters.</p><p><strong>Results: </strong>We discerned 18 clusters consistent with normal appearing tissue (NAT) types and 14 potential ischemic lesion (stage) types, providing insights into the variability and aggressiveness of lesion progression and current anomalous stroke-related imaging features. Of the 529 ischemic stroke lesion instances previously identified by two radiologists, 493 (92 %) were autonomously identified; 460 (87 %) were identified as efficaciously or better than the radiologists.</p><p><strong>Conclusions: </strong>The data analyzed included a small number of clinical patients without follow-up or contemporaneous histology; therefor, the findings and theorizing should be treated as conjecture. Nevertheless, each identified NAT and lesion type is consistent with the known underpinnings of physiological tissues and pathological ischemic stroke lesion (stage) types. Several findings should be considered in current clinical imaging: WM fluid accumulation, BBB compromise conundrum, b<sub>1000</sub> identified core may not be dead tissue, and a practical reason for DWI (pseudo) normalization.</p>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":" ","pages":"110294"},"PeriodicalIF":2.1,"publicationDate":"2024-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142786087","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-02DOI: 10.1016/j.mri.2024.110290
Julia Diamandi, Christian Raimondo, Mahdi Alizadeh, Adam Flanders, Stavropoula Tjoumakaris, M Reid Gooch, Pascal Jabbour, Robert Rosenwasser, Nikolaos Mouchtouris
Purpose: To evaluate the Mean Apparent Propagator (MAP) MRI for processing multi-shell diffusion imaging in patients with acute ischemic stroke (AIS) and correlate to diffusion tensor imaging (DTI) and neurite orientation and dispersion density imaging (NODDI).
Methods: We enrolled patients with AIS from 1/2022 to 4/2024 who underwent multi-shell diffusion imaging on a 3.0-Tesla scanner to generate DTI, NODDI and MAP measures. Mean intensity and standard deviation (SD) were calculated for the infarcted regions-of-interest in b0, fractional anisotropy (FA), mean diffusivity (MD), intra-cellular volume fraction (ICVF), free water fraction (FWF), and orientation dispersion index (ODI), return to the origin probability (RTOP), return to the plane probability (RTPP), return to the axis probability (RTAP), propagator anisotropy (PA), q-space Mean Square Displacement (QMSD), and non-Gaussianity (NG).
Results: Twenty-two patients were included with an average age of 69.5 ± 13.5, mean NIHSS of 12.4 ± 7.7, and median infarct of 73.3 ± 10.1 ml. ICVF was correlated with RTPP (ρ = 0.82, p < 0.01), RTAP (ρ = 0.76, p < 0.01) and RTOP (ρ = 0.79, p < 0.01), ODI with PA (ρ = -0.83, p < 0.01), FWF with RTOP (ρ = -0.73, p < 0.01), RTAP (ρ = -0.69, p < 0.01), and RTPP (ρ = -0.73, p < 0.01), MD with RTPP (ρ = -0.80, p < 0.01), RTOP (ρ = -0.79, p < 0.01), and RTAP (ρ = -0.77, p < 0.01), FA with RTAP (ρ = 0.77, p < 0.01), RTOP (ρ = 0.67, p = 0.01), PA (ρ = 0.74, p < 0.01), and SD PA (ρ = 0.85, p < 0.01). Multivariable linear regression identified the SD QMSD (β = 0.406, p = 0.008), thrombectomy (β = 0.481, p = 0.002), and infarct volume (β = 0.292, p = 0.051) as predictive of stroke severity based on NIHSS.
Conclusions: Given its short processing time, MAP MRI is a valuable alternative with potential for clinical use in AIS.
{"title":"Use of mean apparent propagator (MAP) MRI in patients with acute ischemic stroke: A comparative study with DTI and NODDI.","authors":"Julia Diamandi, Christian Raimondo, Mahdi Alizadeh, Adam Flanders, Stavropoula Tjoumakaris, M Reid Gooch, Pascal Jabbour, Robert Rosenwasser, Nikolaos Mouchtouris","doi":"10.1016/j.mri.2024.110290","DOIUrl":"10.1016/j.mri.2024.110290","url":null,"abstract":"<p><strong>Purpose: </strong>To evaluate the Mean Apparent Propagator (MAP) MRI for processing multi-shell diffusion imaging in patients with acute ischemic stroke (AIS) and correlate to diffusion tensor imaging (DTI) and neurite orientation and dispersion density imaging (NODDI).</p><p><strong>Methods: </strong>We enrolled patients with AIS from 1/2022 to 4/2024 who underwent multi-shell diffusion imaging on a 3.0-Tesla scanner to generate DTI, NODDI and MAP measures. Mean intensity and standard deviation (SD) were calculated for the infarcted regions-of-interest in b0, fractional anisotropy (FA), mean diffusivity (MD), intra-cellular volume fraction (ICVF), free water fraction (FWF), and orientation dispersion index (ODI), return to the origin probability (RTOP), return to the plane probability (RTPP), return to the axis probability (RTAP), propagator anisotropy (PA), q-space Mean Square Displacement (QMSD), and non-Gaussianity (NG).</p><p><strong>Results: </strong>Twenty-two patients were included with an average age of 69.5 ± 13.5, mean NIHSS of 12.4 ± 7.7, and median infarct of 73.3 ± 10.1 ml. ICVF was correlated with RTPP (ρ = 0.82, p < 0.01), RTAP (ρ = 0.76, p < 0.01) and RTOP (ρ = 0.79, p < 0.01), ODI with PA (ρ = -0.83, p < 0.01), FWF with RTOP (ρ = -0.73, p < 0.01), RTAP (ρ = -0.69, p < 0.01), and RTPP (ρ = -0.73, p < 0.01), MD with RTPP (ρ = -0.80, p < 0.01), RTOP (ρ = -0.79, p < 0.01), and RTAP (ρ = -0.77, p < 0.01), FA with RTAP (ρ = 0.77, p < 0.01), RTOP (ρ = 0.67, p = 0.01), PA (ρ = 0.74, p < 0.01), and SD PA (ρ = 0.85, p < 0.01). Multivariable linear regression identified the SD QMSD (β = 0.406, p = 0.008), thrombectomy (β = 0.481, p = 0.002), and infarct volume (β = 0.292, p = 0.051) as predictive of stroke severity based on NIHSS.</p><p><strong>Conclusions: </strong>Given its short processing time, MAP MRI is a valuable alternative with potential for clinical use in AIS.</p>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":" ","pages":"110290"},"PeriodicalIF":2.1,"publicationDate":"2024-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142780478","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-28DOI: 10.1016/j.mri.2024.110283
Ruqi Ou, Yongjun Peng
Endometrial cancer is a common disease in women. Stratifying the risk of early-stage endometrial cancer can aid in personalized treatment for patients. Risk stratification is primarily based on tumor grade, histological type, lymph node metastasis, and depth of myometrial invasion. Multimodal magnetic resonance functional imaging (including DCE-MRI, DWI, IVIM, DTI, DKI) has significant value in assessing the extent of myometrial and cervical infiltration, extrauterine involvement range, determining lymph node metastasis and tumor size. This article provides a brief overview of these techniques.
{"title":"Preoperative risk stratification of early-stage endometrial cancer assessed by multimodal magnetic resonance functional imaging.","authors":"Ruqi Ou, Yongjun Peng","doi":"10.1016/j.mri.2024.110283","DOIUrl":"https://doi.org/10.1016/j.mri.2024.110283","url":null,"abstract":"<p><p>Endometrial cancer is a common disease in women. Stratifying the risk of early-stage endometrial cancer can aid in personalized treatment for patients. Risk stratification is primarily based on tumor grade, histological type, lymph node metastasis, and depth of myometrial invasion. Multimodal magnetic resonance functional imaging (including DCE-MRI, DWI, IVIM, DTI, DKI) has significant value in assessing the extent of myometrial and cervical infiltration, extrauterine involvement range, determining lymph node metastasis and tumor size. This article provides a brief overview of these techniques.</p>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"117 ","pages":"110283"},"PeriodicalIF":2.1,"publicationDate":"2024-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142769985","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-26DOI: 10.1016/j.mri.2024.110282
Chongshuang Yang , Hasyma Abu Hassan , Nur Farhayu Omar , Tze Hui Soo , Ahmad Shuib Bin Yahaya , Tianliang Shi , Zhihong Qin , Min Wu , Jing Yang
Objective
To explore the value of amide proton transfer (APT) imaging in assessing parametrial invasion (PMI) and lymph-vascular space invasion (LVSI) of cervical cancer.
Materials and methods
We retrospectively analyzed the clinical and imaging data of cervical cancer patients diagnosed pathologically at our hospital from January 2021 to June 2024. All patients underwent routine magnetic resonance imaging (MRI), diffusion-weighted imaging (DWI), and APT imaging before treatment. Apparent diffusion coefficient (ADC) and APT values were measured. Based on the pathological results, patients were categorized into LVSI (+) and LVSI (−) groups, and PMI (+) and PMI (−) groups. Independent sample t-tests were used to compare the ADC and APT values between these groups. Receiver operating characteristic (ROC) curves were used to assess the sensitivity, specificity, and area under the curve (AUC) of ADC, APT, and ADC + APT in predicting PMI and LVSI. The Delong test was employed to compare the diagnostic performance among these measures.
Results
A total of 83 patients were included, with 56 in the LVSI (−) group, 27 in the LVSI (+) group, 35 in the PMI (−) group, and 16 in the PMI (+) group. The ADC values for the LVSI (+) and PMI (+) groups were significantly lower than those for the LVSI (−) and PMI (−) groups (P < 0.01). The APT values for the LVSI (+) and PMI (+) groups were significantly higher than those for the LVSI (−) and PMI (−) groups (P < 0.01). The AUC values for ADC, APT, and the combination of ADC + APT in predicting LVSI were 0.839, 0.788, and 0.880, respectively, and in predicting PMI were 0.770, 0.764, and 0.796, respectively. There were no statistically significant differences in the diagnostic performance of ADC, APT, and ADC + APT in predicting PMI. However, the diagnostic performance of ADC + APT in predicting LVSI was significantly better than that of ADC and APT alone (P < 0.01).
Conclusion
APT imaging can predict LVSI and PMI status in cervical cancer before surgery. When combined with ADC, its diagnostic accuracy for predicting LVSI is higher than that of APT or ADC alone. This suggests a novel approach for assessing LVSI in cervical cancer.
{"title":"The value of amide proton transfer imaging in predicting parametrial invasion and lymph-vascular space invasion of cervical cancer","authors":"Chongshuang Yang , Hasyma Abu Hassan , Nur Farhayu Omar , Tze Hui Soo , Ahmad Shuib Bin Yahaya , Tianliang Shi , Zhihong Qin , Min Wu , Jing Yang","doi":"10.1016/j.mri.2024.110282","DOIUrl":"10.1016/j.mri.2024.110282","url":null,"abstract":"<div><h3>Objective</h3><div>To explore the value of amide proton transfer (APT) imaging in assessing parametrial invasion (PMI) and lymph-vascular space invasion (LVSI) of cervical cancer.</div></div><div><h3>Materials and methods</h3><div>We retrospectively analyzed the clinical and imaging data of cervical cancer patients diagnosed pathologically at our hospital from January 2021 to June 2024. All patients underwent routine magnetic resonance imaging (MRI), diffusion-weighted imaging (DWI), and APT imaging before treatment. Apparent diffusion coefficient (ADC) and APT values were measured. Based on the pathological results, patients were categorized into LVSI (+) and LVSI (−) groups, and PMI (+) and PMI (−) groups. Independent sample <em>t</em>-tests were used to compare the ADC and APT values between these groups. Receiver operating characteristic (ROC) curves were used to assess the sensitivity, specificity, and area under the curve (AUC) of ADC, APT, and ADC + APT in predicting PMI and LVSI. The Delong test was employed to compare the diagnostic performance among these measures.</div></div><div><h3>Results</h3><div>A total of 83 patients were included, with 56 in the LVSI (−) group, 27 in the LVSI (+) group, 35 in the PMI (−) group, and 16 in the PMI (+) group. The ADC values for the LVSI (+) and PMI (+) groups were significantly lower than those for the LVSI (−) and PMI (−) groups (<em>P</em> < 0.01). The APT values for the LVSI (+) and PMI (+) groups were significantly higher than those for the LVSI (−) and PMI (−) groups (<em>P</em> < 0.01). The AUC values for ADC, APT, and the combination of ADC + APT in predicting LVSI were 0.839, 0.788, and 0.880, respectively, and in predicting PMI were 0.770, 0.764, and 0.796, respectively. There were no statistically significant differences in the diagnostic performance of ADC, APT, and ADC + APT in predicting PMI. However, the diagnostic performance of ADC + APT in predicting LVSI was significantly better than that of ADC and APT alone (<em>P</em> < 0.01).</div></div><div><h3>Conclusion</h3><div>APT imaging can predict LVSI and PMI status in cervical cancer before surgery. When combined with ADC, its diagnostic accuracy for predicting LVSI is higher than that of APT or ADC alone. This suggests a novel approach for assessing LVSI in cervical cancer.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"116 ","pages":"Article 110282"},"PeriodicalIF":2.1,"publicationDate":"2024-11-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142739838","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-22DOI: 10.1016/j.mri.2024.110278
Wanyu Bian , Albert Jang , Fang Liu
Using single-task deep learning methods to reconstruct Magnetic Resonance Imaging (MRI) data acquired with different imaging sequences is inherently challenging. The trained deep learning model typically lacks generalizability, and the dissimilarity among image datasets with different types of contrast leads to suboptimal learning performance.
This paper proposes a meta-learning approach to efficiently learn image features from multiple MRI datasets. Our algorithm can perform multi-task learning to simultaneously reconstruct MRI images acquired using different imaging sequences with various image contrasts. We have developed a proximal gradient descent-inspired optimization method to learn image features across image and k-space domains, ensuring high-performance learning for every image contrast. Meanwhile, meta-learning, a “learning-to-learn” process, is incorporated into our framework to improve the learning of mutual features embedded in multiple image contrasts.
The experimental results reveal that our proposed multi-task meta-learning approach surpasses state-of-the-art single-task learning methods at high acceleration rates. Our meta-learning consistently delivers accurate and detailed reconstructions, achieves the lowest pixel-wise errors, and significantly enhances qualitative performance across all tested acceleration rates.
We have demonstrated the ability of our new meta-learning reconstruction method to successfully reconstruct highly-undersampled k-space data from multiple MRI datasets simultaneously, outperforming other compelling reconstruction methods previously developed for single-task learning.
使用单任务深度学习方法重建以不同成像序列获取的磁共振成像（MRI）数据本身就具有挑战性。训练好的深度学习模型通常缺乏普适性，不同对比度类型的图像数据集之间的不相似性导致学习性能不理想。本文提出了一种元学习方法，可从多个核磁共振成像数据集中高效学习图像特征。我们的算法可以执行多任务学习，同时重建使用不同成像序列获取的具有不同图像对比度的 MRI 图像。我们开发了一种受近梯度下降启发的优化方法，用于跨图像和 k 空间域学习图像特征，确保对每种图像对比度进行高性能学习。同时，元学习（一种 "从学习到学习 "的过程）被纳入到我们的框架中，以改善嵌入在多种图像对比中的相互特征的学习。实验结果表明，我们提出的多任务元学习方法以较高的加速度超越了最先进的单任务学习方法。在所有测试的加速度下，我们的元学习方法都能持续提供准确、详细的重建，实现最低的像素误差，并显著提高质量性能。我们已经证明，我们新的元学习重建方法能够同时从多个核磁共振成像数据集成功重建高度去采样的 k 空间数据，优于之前为单任务学习开发的其他引人注目的重建方法。
{"title":"Multi-task magnetic resonance imaging reconstruction using meta-learning","authors":"Wanyu Bian , Albert Jang , Fang Liu","doi":"10.1016/j.mri.2024.110278","DOIUrl":"10.1016/j.mri.2024.110278","url":null,"abstract":"<div><div>Using single-task deep learning methods to reconstruct Magnetic Resonance Imaging (MRI) data acquired with different imaging sequences is inherently challenging. The trained deep learning model typically lacks generalizability, and the dissimilarity among image datasets with different types of contrast leads to suboptimal learning performance.</div><div>This paper proposes a meta-learning approach to efficiently learn image features from multiple MRI datasets. Our algorithm can perform multi-task learning to simultaneously reconstruct MRI images acquired using different imaging sequences with various image contrasts. We have developed a proximal gradient descent-inspired optimization method to learn image features across image and k-space domains, ensuring high-performance learning for every image contrast. Meanwhile, meta-learning, a “learning-to-learn” process, is incorporated into our framework to improve the learning of mutual features embedded in multiple image contrasts.</div><div>The experimental results reveal that our proposed multi-task meta-learning approach surpasses state-of-the-art single-task learning methods at high acceleration rates. Our meta-learning consistently delivers accurate and detailed reconstructions, achieves the lowest pixel-wise errors, and significantly enhances qualitative performance across all tested acceleration rates.</div><div>We have demonstrated the ability of our new meta-learning reconstruction method to successfully reconstruct highly-undersampled k-space data from multiple MRI datasets simultaneously, outperforming other compelling reconstruction methods previously developed for single-task learning.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"116 ","pages":"Article 110278"},"PeriodicalIF":2.1,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142695521","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-19DOI: 10.1016/j.mri.2024.110277
Uten Yarach , Itthi Chatnuntawech , Congyu Liao , Surat Teerapittayanon , Siddharth Srinivasan Iyer , Tae Hyung Kim , Justin Haldar , Jaejin Cho , Berkin Bilgic , Yuxin Hu , Brian Hargreaves , Kawin Setsompop
Purpose: BUDA-cEPI has been shown to achieve high-quality, high-resolution diffusion magnetic resonance imaging (dMRI) with fast acquisition time, particularly when used in conjunction with S-LORAKS reconstruction. However, this comes at a cost of more complex reconstruction that is computationally prohibitive. In this work we develop rapid reconstruction pipeline for BUDA-cEPI to pave the way for its deployment in routine clinical and neuroscientific applications. The proposed reconstruction includes the development of ML-based unrolled reconstruction as well as rapid ML-based B0 and eddy current estimations that are needed. The architecture of the unroll network was designed so that it can mimic S-LORAKS regularization well, with the addition of virtual coil channels.
Methods: BUDA-cEPI RUN-UP – a model-based framework that incorporates off-resonance and eddy current effects was unrolled through an artificial neural network with only six gradient updates. The unrolled network alternates between data consistency (i.e., forward BUDA-cEPI and its adjoint) and regularization steps where U-Net plays a role as the regularizer. To handle the partial Fourier effect, the virtual coil concept was also introduced into the reconstruction to effectively take advantage of the smooth phase prior and trained to predict the ground-truth images obtained by BUDA-cEPI with S-LORAKS.
Results: The introduction of the Virtual Coil concept into the unrolled network was shown to be key to achieving high-quality reconstruction for BUDA-cEPI. With the inclusion of an additional non-diffusion image (b-value = 0 s/mm2), a slight improvement was observed, with the normalized root mean square error further reduced by approximately 5 %. The reconstruction times for S-LORAKS and the proposed unrolled networks were approximately 225 and 3 s per slice, respectively.
Conclusion: BUDA-cEPI RUN-UP was shown to reduce the reconstruction time by ∼88× when compared to the state-of-the-art technique, while preserving imaging details as demonstrated through DTI application.
{"title":"Blip-up blip-down circular EPI (BUDA-cEPI) for distortion-free dMRI with rapid unrolled deep learning reconstruction","authors":"Uten Yarach , Itthi Chatnuntawech , Congyu Liao , Surat Teerapittayanon , Siddharth Srinivasan Iyer , Tae Hyung Kim , Justin Haldar , Jaejin Cho , Berkin Bilgic , Yuxin Hu , Brian Hargreaves , Kawin Setsompop","doi":"10.1016/j.mri.2024.110277","DOIUrl":"10.1016/j.mri.2024.110277","url":null,"abstract":"<div><div>Purpose: BUDA-cEPI has been shown to achieve high-quality, high-resolution diffusion magnetic resonance imaging (dMRI) with fast acquisition time, particularly when used in conjunction with S-LORAKS reconstruction. However, this comes at a cost of more complex reconstruction that is computationally prohibitive. In this work we develop rapid reconstruction pipeline for BUDA-cEPI to pave the way for its deployment in routine clinical and neuroscientific applications. The proposed reconstruction includes the development of ML-based unrolled reconstruction as well as rapid ML-based B0 and eddy current estimations that are needed. The architecture of the unroll network was designed so that it can mimic S-LORAKS regularization well, with the addition of virtual coil channels.</div><div>Methods: BUDA-cEPI RUN-UP – a model-based framework that incorporates off-resonance and eddy current effects was unrolled through an artificial neural network with only six gradient updates. The unrolled network alternates between data consistency (i.e., forward BUDA-cEPI and its adjoint) and regularization steps where U-Net plays a role as the regularizer. To handle the partial Fourier effect, the virtual coil concept was also introduced into the reconstruction to effectively take advantage of the smooth phase prior and trained to predict the ground-truth images obtained by BUDA-cEPI with S-LORAKS.</div><div>Results: The introduction of the Virtual Coil concept into the unrolled network was shown to be key to achieving high-quality reconstruction for BUDA-cEPI. With the inclusion of an additional non-diffusion image (b-value = 0 s/mm<sup>2</sup>), a slight improvement was observed, with the normalized root mean square error further reduced by approximately 5 %. The reconstruction times for S-LORAKS and the proposed unrolled networks were approximately 225 and 3 s per slice, respectively.</div><div>Conclusion: BUDA-cEPI RUN-UP was shown to reduce the reconstruction time by ∼88× when compared to the state-of-the-art technique, while preserving imaging details as demonstrated through DTI application.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"115 ","pages":"Article 110277"},"PeriodicalIF":2.1,"publicationDate":"2024-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142682124","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-19DOI: 10.1016/j.mri.2024.110276
Lu Yang , Longwu Yu , Guangzi Shi , Lingjie Yang , Yu Wang , Riyu Han , Fengqiong Huang , Yinfeng Qian , Xiaohui Duan
Purpose
This study aimed to investigate the potential of radiomic features derived from dynamic contrast-enhanced MRI (DCE-MRI) in predicting Ki-67 and p16 status in head and neck squamous cell carcinoma (HNSCC).
Materials and methods
A cohort of 124 HNSCC patients who underwent pre-surgery DCE-MRI were included and divided into training and test set (7:3), further subgroup analysis was performed for 104 cases with oral squamous cell carcinoma (OSCC). Radiomics features were extracted from DCE images. The least absolute shrinkage and selection operator (LASSO) was used for radiomics features selection, and receiver operating characteristics analysis for predictive performance assessment. The nomogram's performance was evaluated using decision curve analysis (DCA).
Results
Ten DCE-MRI features were identified to build the predictive model of HNSCC, demonstrating excellent predictive value for Ki-67 status in both the training set (AUC of 0.943) and test set (AUC of 0.801). The nomograms based on the predictive model showed good fit in the calibration curves (p > 0.05), and DCA indicated its high clinical usefulness. In subgroup analysis of OSCC, fourteen features were selected to build the predictive model for Ki-67 status with an AUC of 0.960 in training set and 0.817 in test set. No features could be included to establish a model to predict p16 status.
Conclusion
The radiomics model utilizing DCE-MRI features could effectively predict Ki-67 status in HNSCC patients, offering potential for noninvasive preoperative prediction of Ki-67 status.
{"title":"Radiomic features of dynamic contrast-enhanced MRI can predict Ki-67 status in head and neck squamous cell carcinoma","authors":"Lu Yang , Longwu Yu , Guangzi Shi , Lingjie Yang , Yu Wang , Riyu Han , Fengqiong Huang , Yinfeng Qian , Xiaohui Duan","doi":"10.1016/j.mri.2024.110276","DOIUrl":"10.1016/j.mri.2024.110276","url":null,"abstract":"<div><h3>Purpose</h3><div>This study aimed to investigate the potential of radiomic features derived from dynamic contrast-enhanced MRI (DCE-MRI) in predicting Ki-67 and p16 status in head and neck squamous cell carcinoma (HNSCC).</div></div><div><h3>Materials and methods</h3><div>A cohort of 124 HNSCC patients who underwent pre-surgery DCE-MRI were included and divided into training and test set (7:3), further subgroup analysis was performed for 104 cases with oral squamous cell carcinoma (OSCC). Radiomics features were extracted from DCE images. The least absolute shrinkage and selection operator (LASSO) was used for radiomics features selection, and receiver operating characteristics analysis for predictive performance assessment. The nomogram's performance was evaluated using decision curve analysis (DCA).</div></div><div><h3>Results</h3><div>Ten DCE-MRI features were identified to build the predictive model of HNSCC, demonstrating excellent predictive value for Ki-67 status in both the training set (AUC of 0.943) and test set (AUC of 0.801). The nomograms based on the predictive model showed good fit in the calibration curves (<em>p</em> > 0.05), and DCA indicated its high clinical usefulness. In subgroup analysis of OSCC, fourteen features were selected to build the predictive model for Ki-67 status with an AUC of 0.960 in training set and 0.817 in test set. No features could be included to establish a model to predict p16 status.</div></div><div><h3>Conclusion</h3><div>The radiomics model utilizing DCE-MRI features could effectively predict Ki-67 status in HNSCC patients, offering potential for noninvasive preoperative prediction of Ki-67 status.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"116 ","pages":"Article 110276"},"PeriodicalIF":2.1,"publicationDate":"2024-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142687383","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-17DOI: 10.1016/j.mri.2024.110279
Shahzad Ahmed , Feng Jinchao , Javed Ferzund , Muhammad Usman Ali , Muhammad Yaqub , Malik Abdul Manan , Atif Mehmood
Purpose
This study introduces GraFMRI, a novel framework designed to address the challenges of reconstructing high-quality MRI images from undersampled k-space data. Traditional methods often suffer from noise amplification and loss of structural detail, leading to suboptimal image quality. GraFMRI leverages Graph Neural Networks (GNNs) to transform multi-modal MRI data (T1, T2, PD) into a graph-based representation, enabling the model to capture intricate spatial relationships and inter-modality dependencies.
Methods
The framework integrates Graph-Based Non-Local Means (NLM) Filtering for effective noise suppression and Adversarial Training to reduce artifacts. A dynamic attention mechanism enables the model to focus on key anatomical regions, even when fully-sampled reference images are unavailable. GraFMRI was evaluated on the IXI and fastMRI datasets using Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) as metrics for reconstruction quality.
Results
GraFMRI consistently outperforms traditional and self-supervised reconstruction techniques. Significant improvements in multi-modal fusion were observed, with better preservation of information across modalities. Noise suppression through NLM filtering and artifact reduction via adversarial training led to higher PSNR and SSIM scores across both datasets. The dynamic attention mechanism further enhanced the accuracy of the reconstructions by focusing on critical anatomical regions.
Conclusion
GraFMRI provides a scalable, robust solution for multi-modal MRI reconstruction, addressing noise and artifact challenges while enhancing diagnostic accuracy. Its ability to fuse information from different MRI modalities makes it adaptable to various clinical applications, improving the quality and reliability of reconstructed images.
{"title":"GraFMRI: A graph-based fusion framework for robust multi-modal MRI reconstruction","authors":"Shahzad Ahmed , Feng Jinchao , Javed Ferzund , Muhammad Usman Ali , Muhammad Yaqub , Malik Abdul Manan , Atif Mehmood","doi":"10.1016/j.mri.2024.110279","DOIUrl":"10.1016/j.mri.2024.110279","url":null,"abstract":"<div><h3>Purpose</h3><div>This study introduces GraFMRI, a novel framework designed to address the challenges of reconstructing high-quality MRI images from undersampled k-space data. Traditional methods often suffer from noise amplification and loss of structural detail, leading to suboptimal image quality. GraFMRI leverages Graph Neural Networks (GNNs) to transform multi-modal MRI data (T1, T2, PD) into a graph-based representation, enabling the model to capture intricate spatial relationships and inter-modality dependencies.</div></div><div><h3>Methods</h3><div>The framework integrates Graph-Based Non-Local Means (NLM) Filtering for effective noise suppression and Adversarial Training to reduce artifacts. A dynamic attention mechanism enables the model to focus on key anatomical regions, even when fully-sampled reference images are unavailable. GraFMRI was evaluated on the IXI and fastMRI datasets using Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) as metrics for reconstruction quality.</div></div><div><h3>Results</h3><div>GraFMRI consistently outperforms traditional and self-supervised reconstruction techniques. Significant improvements in multi-modal fusion were observed, with better preservation of information across modalities. Noise suppression through NLM filtering and artifact reduction via adversarial training led to higher PSNR and SSIM scores across both datasets. The dynamic attention mechanism further enhanced the accuracy of the reconstructions by focusing on critical anatomical regions.</div></div><div><h3>Conclusion</h3><div>GraFMRI provides a scalable, robust solution for multi-modal MRI reconstruction, addressing noise and artifact challenges while enhancing diagnostic accuracy. Its ability to fuse information from different MRI modalities makes it adaptable to various clinical applications, improving the quality and reliability of reconstructed images.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"116 ","pages":"Article 110279"},"PeriodicalIF":2.1,"publicationDate":"2024-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142676371","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-17DOI: 10.1016/j.mri.2024.110273
John C. Gore
There has been tremendous progress in MRI over the past 40+ years, driven by advances in technology as well as human ingenuity, with considerable impact in medicine. However, our understanding of how to account for, and interpret, MRI properties quantitatively lags behind these technical advances. This lack of understanding will limit our ability to make full use of quantitative metrics in the future, and much more work is needed to bridge this knowledge gap.
{"title":"Progress in MRI is NOT ubiquitous","authors":"John C. Gore","doi":"10.1016/j.mri.2024.110273","DOIUrl":"10.1016/j.mri.2024.110273","url":null,"abstract":"<div><div>There has been tremendous progress in MRI over the past 40+ years, driven by advances in technology as well as human ingenuity, with considerable impact in medicine. However, our understanding of how to account for, and interpret, MRI properties <em>quantitatively</em> lags behind these technical advances. This lack of understanding will limit our ability to make full use of quantitative metrics in the future, and much more work is needed to bridge this knowledge gap.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"115 ","pages":"Article 110273"},"PeriodicalIF":2.1,"publicationDate":"2024-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142648472","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-16DOI: 10.1016/j.mri.2024.110280
Teodoro Martín-Noguerol , Pilar López-Úbeda , Félix Paulano-Godino , Antonio Luna
In this letter to the editor, authors highlight the key role of data labeling in training AI models for medical imaging, discussing the complexities, resource demands, costs, and the relevance of quality control in the labeling process including the potential and limitations of AI tools for automated labeling. The article underlines that labeling quality is essential for the accuracy of AI models and the safety of their clinical applications, highlighting the legal responsibilities of labelers in cases where improper labeling leads to AI errors.
{"title":"Manual data labeling, radiology, and artificial intelligence: It is a dirty job, but someone has to do it","authors":"Teodoro Martín-Noguerol , Pilar López-Úbeda , Félix Paulano-Godino , Antonio Luna","doi":"10.1016/j.mri.2024.110280","DOIUrl":"10.1016/j.mri.2024.110280","url":null,"abstract":"<div><div>In this letter to the editor, authors highlight the key role of data labeling in training AI models for medical imaging, discussing the complexities, resource demands, costs, and the relevance of quality control in the labeling process including the potential and limitations of AI tools for automated labeling. The article underlines that labeling quality is essential for the accuracy of AI models and the safety of their clinical applications, highlighting the legal responsibilities of labelers in cases where improper labeling leads to AI errors.</div></div>","PeriodicalId":18165,"journal":{"name":"Magnetic resonance imaging","volume":"116 ","pages":"Article 110280"},"PeriodicalIF":2.1,"publicationDate":"2024-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142667820","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号