MRI Volumes Research Articles

Abstract PS05-07: Early prediction of response to Neoadjuvant Immunotherapy in Triple Negative Breast Cancer (TNBC) with DCE-MRI

Abstract Purpose: Neoadjuvant immunotherapy (NIT) in combination with neoadjuvant chemotherapy (NCT) was recently approved for treatment of TNBC patients with increased rates of pathologic complete response (pCR) compared to NCT alone. The aim of this study was to evaluate if dynamic contrast-enhanced (DCE)-MRI performed after 2 and/or 4 cycles of NIT + NCT, can predict which patients will achieve pCR, potentially triaging them to continuation of NIT+NCT or, when appropriate, to de-escalation trials. Alternatively, identified chemoresistant tumors who are unlikely to achieve pCR may be directed to other treatment strategies, including novel targeted trials, and avoid the unnecessary toxicity of NIT. Methods and Materials: Preliminary analysis included 64 patients from prospective IRB-approved study (NCT02276443) with stage I-III TNBC who underwent DCE-MRI at baseline (BL), after 2 cycles (C2), and 4 cycles (C4) of NIT combined with standard of care NCT (Paclitaxel +/- carboplatin). Tumor volumes were calculated using 3 axis measurements of the index lesion at the DCE MRI and percent tumor volume reduction (TVR) between BL, C2, and C4 was calculated. pCR was assessed at surgery after completion of neoadjuvant treatment. Correlation between pCR and TVR was evaluated using ROC analysis. Results: 59% (38/64) of TNBC patients achieved pCR after NIT+NCT. DCE-MRI after 2 cycles of NIT+NCT was able to predict pCR with an AUC of 0.71 (95% CI: 0.57-0.84). TVR >90% at C2 predicted pCR with PPV 86%, and TVR < 35% predicted chemoresistance with NPV 100%. Following 4 cycles of treatment DCE-MRI was able to predict pCR with an AUC of 0.81 (95% CI: 0.69-0.92). TVR >95% at C4 was predictive of chemosensitivity with PPV 82%, while TVR < 75% was predictive of chemoresistance with NPV 100%. Conclusions: DCE-MRI volumetric changes early during NIT + NCT were able to predict pCR status of TNBC patients as either excellent responders or nonresponders, triaging them to SOC neoadjuvant therapy with option for de-escalation trials, or targeted therapies, respectively. These preliminary results will be validated in the larger cohort after completion of the ongoing prospective clinical trial. Citation Format: Gaiane Rauch, Mary Guirguis, Miral Patel, Rosalind Candelaria, Rania Mohamed, Tanya Moseley, H. T. Carisa Le-Petross, Jessica Leung, Gary Whitman, Deanna Lane, Marion Scoggins, Frances Perez, Jia Sun, Sanaz Pashapoor, Zhan Xu, Jason White, Peng Wei, Brandy Reed, Jong Bum Son, Ken-Pin Hwang, Bikash Panthi, Anil Korkut, Lei Huo, Kelly Hunt, Alyson Clayborn, Jennifer Litton, Vicente Valero, Debu Tripathy, Clinton Yam, Wei Yang, Jingfei Ma, Beatriz Adrada. Early prediction of response to Neoadjuvant Immunotherapy in Triple Negative Breast Cancer (TNBC) with DCE-MRI [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PS05-07.

3D‐SRDM: 3D super‐resolution of MRI volumes based on diffusion model

AbstractBrain magnetic resonance imaging (MRI) is crucial for diagnosing and understanding neurological disorders, but inherent limitations hinder the visualization of fine details in brain structures. The emergence of super‐resolution techniques, especially deep‐learning methods, has improved imaging quality of MRI, by increasing MRI spatial resolution. At present, deep‐learning algorithms mostly performed super‐resolution on 2D MRI images. However, considering 3D nature of MRI, 3D models are more suitable for brain MRI super‐resolution. To achieve finer brain structural details, this study proposes a 3D brain MRI super‐resolution method based on diffusion model (3D‐SRDM), which is a fast and easily trainable neural network for the generation of high‐resolution brain MRI images. In our 3D‐SRDM model, the self‐attention module in U‐Net is replaced with 3D spatial attention mechanism. The network structure of 3D‐SRDM is optimized to reduce training parameters. Moreover, accelerated sampling from denoising diffusion implicit model is also incorporated to reduce time consumption. By these optimizations, compared with original diffusion model, the proposed model can achieve about 10‐ and 5‐fold speed increase at 4× and 8× super‐resolution of 3D brain MRI volumes, respectively, almost without affecting image quality. Thus, 3D‐SRDM has potential application value in efficiently generating high‐resolution 3D brain MRI images, thus facilitating the doctors' diagnosis.

MRI-based radiomics for predicting histology in malignant salivary gland tumors: methodology and "proof of principle".

Defining the exact histological features of salivary gland malignancies before treatment remains an unsolved problem that compromises the ability to tailor further therapeutic steps individually. Radiomics, a new methodology to extract quantitative information from medical images, could contribute to characterizing the individual cancer phenotype already before treatment in a fast and non-invasive way. Consequently, the standardization and implementation of radiomic analysis in the clinical routine work to predict histology of salivary gland cancer (SGC) could also provide improvements in clinical decision-making. In this study, we aimed to investigate the potential of radiomic features as imaging biomarker to distinguish between high grade and low-grade salivary gland malignancies. We have also investigated the effect of image and feature level harmonization on the performance of radiomic models. For this study, our dual center cohort consisted of 126 patients, with histologically proven SGC, who underwent curative-intent treatment in two tertiary oncology centers. We extracted and analyzed the radiomics features of 120 pre-therapeutic MRI images with gadolinium (T1 sequences), and correlated those with the definitive post-operative histology. In our study the best radiomic model achieved average AUC of 0.66 and balanced accuracy of 0.63. According to the results, there is significant difference between the performance of models based on MRI intensity normalized images + harmonized features and other models (p value < 0.05) which indicates that in case of dealing with heterogeneous dataset, applying the harmonization methods is beneficial. Among radiomic features minimum intensity from first order, and gray level-variance from texture category were frequently selected during multivariate analysis which indicate the potential of these features as being used as imaging biomarker. The present bicentric study presents for the first time the feasibility of implementing MR-based, handcrafted radiomics, based on T1 contrast-enhanced sequences and the ComBat harmonization method in an effort to predict the formal grading of salivary gland carcinoma with satisfactory performance.

DF-QSM: Data Fidelity based Hybrid Approach for Improved Quantitative Susceptibility Mapping of the Brain.

Quantitative Susceptibility Mapping (QSM) is an advanced magnetic resonance imaging (MRI) technique to quantify the magnetic susceptibility of the tissue under investigation. Deep learning methods have shown promising results in deconvolving the susceptibility distribution from the measured local field obtained from the MR phase. Although existing deep learning based QSM methods can produce high-quality reconstruction, they are highly biased toward training data distribution with less scope for generalizability. This work proposes a hybrid two-step reconstruction approach to improve deep learning based QSM reconstruction. The susceptibility map prediction obtained from the deep learning methods has been refined in the framework developed in this work to ensure consistency with the measured local field. The developed method was validated on existing deep learning and model-based deep learning methods for susceptibility mapping of the brain. The developed method resulted in improved reconstruction for MRI volumes obtained with different acquisition settings, including deep learning models trained on constrained (limited) data settings.

Clinicopathologic Heterogeneity and Glial Activation Patterns in Alzheimer Disease

Factors associated with clinical heterogeneity in Alzheimer disease (AD) lay along a continuum hypothesized to associate with tangle distribution and are relevant for understanding glial activation considerations in therapeutic advancement. To examine clinicopathologic and neuroimaging characteristics of disease heterogeneity in AD along a quantitative continuum using the corticolimbic index (CLix) to account for individuality of spatially distributed tangles found at autopsy. This cross-sectional study was a retrospective medical record review performed on the Florida Autopsied Multiethnic (FLAME) cohort accessioned from 1991 to 2020. Data were analyzed from December 2022 to December 2023. Structural magnetic resonance imaging (MRI) and tau positron emission tomography (PET) were evaluated in an independent neuroimaging group. The FLAME cohort includes 2809 autopsied individuals; included in this study were neuropathologically diagnosed AD cases (FLAME-AD). A digital pathology subgroup of FLAME-AD cases was derived for glial activation analyses. Clinicopathologic factors of heterogeneity that inform patient history and neuropathologic evaluation of AD; CLix score (lower, relative cortical predominance/hippocampal sparing vs higher, relative cortical sparing/limbic predominant cases); neuroimaging measures (ie, structural MRI and tau-PET). Of the 2809 autopsied individuals in the FLAME cohort, 1361 neuropathologically diagnosed AD cases were evaluated. A digital pathology subgroup included 60 FLAME-AD cases. The independent neuroimaging group included 93 cases. Among the 1361 FLAME-AD cases, 633 were male (47%; median [range] age at death, 81 [54-96] years) and 728 were female (53%; median [range] age at death, 81 [53-102] years). A younger symptomatic onset (Spearman ρ = 0.39, P < .001) and faster decline on the Mini-Mental State Examination (Spearman ρ = 0.27; P < .001) correlated with a lower CLix score in FLAME-AD series. Cases with a nonamnestic syndrome had lower CLix scores (median [IQR], 13 [9-18]) vs not (median [IQR], 21 [15-27]; P < .001). Hippocampal MRI volume (Spearman ρ = -0.45; P < .001) and flortaucipir tau-PET uptake in posterior cingulate and precuneus cortex (Spearman ρ = -0.74; P < .001) inversely correlated with CLix score. Although AD cases with a CLix score less than 10 had higher cortical tangle count, we found lower percentage of CD68-activated microglia/macrophage burden (median [IQR], 0.46% [0.32%-0.75%]) compared with cases with a CLix score of 10 to 30 (median [IQR], 0.75% [0.51%-0.98%]) and on par with a CLix score of 30 or greater (median [IQR], 0.40% [0.32%-0.57%]; P = .02). Findings show that AD heterogeneity exists along a continuum of corticolimbic tangle distribution. Reduced CD68 burden may signify an underappreciated association between tau accumulation and microglia/macrophages activation that should be considered in personalized therapy for immune dysregulation.

Unsupervised Multi-parametric MRI Registration Using Neural Optimal Transport.

Precise deformable image registration of multi-parametric MRI sequences is necessary for radiologists in order to identify abnormalities and diagnose diseases, such as prostate cancer and lymphoma. Despite recent advances in unsupervised learning-based registration, volumetric medical image registration that requires considering the variety of data distributions is still challenging. To address the problem of multi-parametric MRI sequence data registration, we propose an unsupervised domain-transported registration method, called OTMorph by employing neural optimal transport that learns an optimal transport plan to map different data distributions. We have designed a novel framework composed of a transport module and a registration module: the former transports data distribution from the moving source domain to the fixed target domain, and the latter takes the transported data and provides the deformed moving volume that is aligned with the fixed volume. Through end-to-end learning, our proposed method can effectively learn deformable registration for the volumes in different distributions. Experimental results with abdominal multi-parametric MRI sequence data show that our method has superior performance over around 67-85% in deforming the MRI volumes compared to the existing learning-based methods. Our method is generic in nature and can be used to register inter-/intra-modality images by mapping the different data distributions in network training.

GAN-Based Motion Artifact Correction of 3D MR Volumes Using an Image-to-Image Translation Algorithm.

The quality of brain MRI volumes is often compromised by motion artifacts arising from intricate respiratory patterns and involuntary head movements, manifesting as blurring and ghosting that markedly degrade imaging quality. In this study, we introduce an innovative approach employing a 3D deep learning framework to restore brain MR volumes afflicted by motion artifacts. The framework integrates a densely connected 3D U-net architecture augmented by generative adversarial network (GAN)-informed training with a novel volumetric reconstruction loss function tailored to 3D GAN to enhance the quality of the volumes. Our methodology is substantiated through comprehensive experimentation involving a diverse set of motion artifact-affected MR volumes. The generated high-quality MR volumes have similar volumetric signatures comparable to motion-free MR volumes after motion correction. This underscores the significant potential of harnessing this 3D deep learning system to aid in the rectification of motion artifacts in brain MR volumes, highlighting a promising avenue for advanced clinical applications.

Automated Liver Volumetry and Hepatic Steatosis Quantification with Magnetic Resonance Imaging Proton Density Fat Fraction

BackgroundPreoperative imaging evaluation of liver volume and hepatic steatosis for the donor affects transplantation outcomes. However, computed tomography (CT) for liver volumetry and magnetic resonance spectroscopy (MRS) for hepatic steatosis are time consuming. Therefore, we investigated the correlation of automated 3D-multi-echo-Dixon sequence magnetic resonance imaging (ME-Dixon MRI) and its derived proton density fat fraction (MRI-PDFF) with CT liver volumetry and MRS hepatic steatosis measurements in living liver donors. MethodsThis retrospective cross-sectional study was conducted from December 2017 to November 2022. We enrolled donors who received a dynamic CT scan and an MRI exam within 2 days. First, the CT volumetry was processed semiautomatically using commercial software, and ME-Dixon MRI volumetry was automatically measured using an embedded sequence. Next, the signal intensity of MRI-PDFF volumetric data was correlated with MRS as the gold standard. ResultsWe included the 165 living donors. The total liver volume of ME-Dixon MRI was significantly correlated with CT (r = 0.913, p < 0.001). The fat percentage measured using MRI-PDFF revealed a strong correlation between automatic segmental volume and MRS (r = 0.705, p < 0.001). Furthermore, the hepatic steatosis group (MRS ≥ 5%) had a strong correlation than the non–hepatic steatosis group (MRS < 5%) in both volumetric (r = 0.906 vs. r = 0.887) and fat fraction analysis (r = 0.779 vs. r = 0.338). ConclusionAutomated ME-Dixon MRI liver volumetry and MRI-PDFF were strongly correlated with CT liver volumetry and MRS hepatic steatosis measurements, especially in donors with hepatic steatosis.

Psychosis in Alzheimer’s disease is associated with specific changes in brain MRI volume, cognition and neuropathology

Psychosis in Alzheimer’s Disease (AD) is prevalent and indicates poor prognosis. However, the neuropathological, cognitive and brain atrophy patterns underlying these symptoms have not been fully elucidated. In this study, we evaluated 178 patients with AD neuropathological change (ADNC) and ante-mortem volumetric brain magnetic resonance imaging (MRI). Presence of psychosis was determined using the Neuropsychiatric Inventory Questionnaire. Clinical Dementia Rating Sum-of-boxes (CDR-SB) was longitudinally compared between groups with a follow-up of 3000 days using mixed-effects multiple linear regression. Neuropsychological tests closest to the time of MRI and brain regional volumes were cross-sectionally compared. Psychosis was associated with lower age of death, higher longitudinal CDR-SB scores, multi-domain cognitive deficits, higher neuritic plaque severity, Braak stage, Lewy Body pathology (LB) and right temporal lobe regional atrophy. Division according to the presence of LB showed differential patterns of AD-typical pathology, cognitive deficits and regional atrophy. In conclusion, psychosis in ADNC with and without LB has clinical value and associates with subgroup patterns of neuropathology, cognition and regional atrophy.

Histologic tau lesions and magnetic resonance imaging biomarkers differ across two progressive supranuclear palsy variants.

Progressive supranuclear palsy is a neurodegenerative disease characterized by the deposition of four-repeat tau in neuronal and glial lesions in the brainstem, cerebellar, subcortical and cortical brain regions. There are varying clinical presentations of progressive supranuclear palsy with different neuroimaging signatures, presumed to be due to different topographical distributions and burden of tau. The classic Richardson syndrome presentation is considered a subcortical variant, whilst progressive supranuclear palsy with predominant speech and language impairment is considered a cortical variant, although the pathological underpinnings of these variants are unclear. In this case-control study, we aimed to determine whether patterns of regional tau pathology differed between these variants and whether tau burden correlated with neuroimaging. Thirty-three neuropathologically confirmed progressive supranuclear palsy patients with either the Richardson syndrome (n = 17) or speech/language (n = 16) variant and ante-mortem magnetic resonance imaging were included. Tau lesion burden was semi-quantitatively graded in cerebellar, brainstem, subcortical and cortical regions and combined to form neuronal and glial tau scores. Regional magnetic resonance imaging volumes were converted to Z-scores using 33 age- and sex-matched controls. Diffusion tensor imaging metrics, including fractional anisotropy and mean diffusivity, were calculated. Tau burden and neuroimaging metrics were compared between groups and correlated using linear regression models. Neuronal and glial tau burden were higher in motor and superior frontal cortices in the speech/language variant. In the subcortical and brainstem regions, only the glial tau burden differed, with a higher burden in globus pallidus, subthalamic nucleus, substantia nigra and red nucleus in Richardson's syndrome. No differences were observed in the cerebellar dentate and striatum. Greater volume loss was observed in the motor cortex in the speech/language variant and in the subthalamic nucleus, red nucleus and midbrain in Richardson's syndrome. Fractional anisotropy was lower in the midbrain and superior cerebellar peduncle in Richardson's syndrome. Mean diffusivity was greater in the superior frontal cortex in the speech/language variant and midbrain in Richardson's syndrome. Neuronal tau burden showed associations with volume loss, lower fractional anisotropy and higher mean diffusivity in the superior frontal cortex, although these findings did not survive correction for multiple comparisons. Results suggest that a shift in the distribution of tau, particularly neuronal tau, within the progressive supranuclear palsy network of regions is driving different clinical presentations in progressive supranuclear palsy. The possibility of different disease epicentres in these clinical variants has potential implications for the use of imaging biomarkers in progressive supranuclear palsy.

Learning Spatio-Temporal Model of Disease Progression With NeuralODEs From Longitudinal Volumetric Data.

Robust forecasting of the future anatomical changes inflicted by an ongoing disease is an extremely challenging task that is out of grasp even for experienced healthcare professionals. Such a capability, however, is of great importance since it can improve patient management by providing information on the speed of disease progression already at the admission stage, or it can enrich the clinical trials with fast progressors and avoid the need for control arms by the means of digital twins. In this work, we develop a deep learning method that models the evolution of age-related disease by processing a single medical scan and providing a segmentation of the target anatomy at a requested future point in time. Our method represents a time-invariant physical process and solves a large-scale problem of modeling temporal pixel-level changes utilizing NeuralODEs. In addition, we demonstrate the approaches to incorporate the prior domain-specific constraints into our method and define temporal Dice loss for learning temporal objectives. To evaluate the applicability of our approach across different age-related diseases and imaging modalities, we developed and tested the proposed method on the datasets with 967 retinal OCT volumes of 100 patients with Geographic Atrophy and 2823 brain MRI volumes of 633 patients with Alzheimer's Disease. For Geographic Atrophy, the proposed method outperformed the related baseline models in the atrophy growth prediction. For Alzheimer's Disease, the proposed method demonstrated remarkable performance in predicting the brain ventricle changes induced by the disease, achieving the state-of-the-art result on TADPOLE cross-sectional prediction challenge dataset.

A Phase 1 Trial of Salvage Stereotactic Body Radiation Therapy for Radiorecurrent Prostate Cancer After Brachytherapy

NCT03253744 is a phase 1 trial with the primary objective to identify the maximum tolerated dose (MTD) of salvage stereotactic body radiation therapy (SBRT) in patients with local prostate cancer recurrence after brachytherapy. Additional objectives included biochemical control and imaging response. This trial was initially designed to test 3 therapeutic dose levels (DLs): 40 Gy (DL1), 42.5 Gy (DL2), and 45 Gy (DL3) in 5 fractions. Intensity modulation was used to deliver the prescription dose to the magnetic resonance imaging and prostate-specific membrane antigen-based positron emission tomography imaging-defined gross tumor volume while simultaneously delivering 30 Gy to an elective volume defined by the prostate gland. This phase 1 trial followed a 3+3 design with a 3-patient expansion at the MTD. Toxicities were scored until trial completion at 2 years post-SBRT using Common Terminology Criteria for Adverse Events version 5.0. Escalation was halted if 2 dose limiting toxicities occurred, defined as any persistent (>4 days) grade 3 toxicity occurring within the first 3 weeks after SBRT or any grade ≥3 genitourinary (GU) or grade 4 gastrointestinal toxicity thereafter. Between August 2018 and January 2023, 9 patients underwent salvage SBRT and were observed for a median of 22 months (Q1-Q3, 20-43 months). No grade 3 to 5 adverse events related to study treatment were observed; thus, no dose limiting toxicities occurred during the observation period. Escalation was halted by amendment given excellent biochemical control in DL1 and DL2 in the setting of a high incidence of clinically significant late grade 2 GU toxicity. Therefore, the MTD was considered 42.5 Gy in 5 fractions (DL2). One- and 2-year biochemical progression-free survival were 100% and 86%, representing a single patient in the trial cohort with biochemical failure (prostate-specific antigen [PSA] nadir + 2.0) at 20 months posttreatment. The MTD of salvage SBRT for the treatment of intraprostatic radiorecurrence after brachytherapy was 42.5 Gy in 5 fractions producing an 86% 2-year biochemical progression-free survival rate, with 1 poststudy failure at 20 months. The most frequent clinically significant toxicity was late grade 2 GU toxicity.

Anatomical correlates of apathy and impulsivity co-occurrence in early Parkinson’s disease

BackgroundAlthough apathy and impulse control disorders (ICDs) are considered to represent opposite extremes of a continuum of motivated behavior (i.e., hypo- and hyperdopaminergic behaviors), they may also co-occur in Parkinson’s disease (PD).ObjectivesWe aimed to explore the co-occurrence of ICDs and apathy and its neural correlates analyzing gray matter (GM) changes in early untreated PD patients. Moreover, we aimed to investigate the possible longitudinal relationship between ICDs and apathy and their putative impact on cognition during the first five years of PD.MethodsWe used the Parkinson’s Progression Markers Initiative (PPMI) database to identify the co-occurrence of apathy and ICDs in 423 early drug-naïve PD patients at baseline and at 5-year follow-up. Baseline MRI volumes and gray matter changes were analyzed between groups using voxel-based morphometry. Multi-level models assessed the longitudinal relationship (across five years) between apathy and ICDs and cognitive functioning.ResultsAt baseline, co-occurrence of apathy and ICDs was observed in 23 patients (5.4%). This finding was related to anatomical GM reduction along the cortical regions involved in the limbic circuit and cognitive control systems. Longitudinal analyses indicated that apathy and ICDs were related to each other as well as to the combined use of levodopa and dopamine agonists. Worse apathetic and ICDs states were associated with poorer executive functions.ConclusionsApathy and ICDs are joint non-exclusive neuropsychiatric disorders also in the early stages of PD and their co-occurrence was associated with GM decrease in several cortical regions of the limbic circuit and cognitive control systems.

Expectation maximisation pseudo labels

In this paper, we study pseudo-labelling. Pseudo-labelling employs raw inferences on unlabelled data as pseudo-labels for self-training. We elucidate the empirical successes of pseudo-labelling by establishing a link between this technique and the Expectation Maximisation algorithm. Through this, we realise that the original pseudo-labelling serves as an empirical estimation of its more comprehensive underlying formulation. Following this insight, we present a full generalisation of pseudo-labels under Bayes’ theorem, termed Bayesian Pseudo Labels. Subsequently, we introduce a variational approach to generate these Bayesian Pseudo Labels, involving the learning of a threshold to automatically select high-quality pseudo labels. In the remainder of the paper, we showcase the applications of pseudo-labelling and its generalised form, Bayesian Pseudo-Labelling, in the semi-supervised segmentation of medical images. Specifically, we focus on: (1) 3D binary segmentation of lung vessels from CT volumes; (2) 2D multi-class segmentation of brain tumours from MRI volumes; (3) 3D binary segmentation of whole brain tumours from MRI volumes; and (4) 3D binary segmentation of prostate from MRI volumes. We further demonstrate that pseudo-labels can enhance the robustness of the learned representations. The code is released in the following GitHub repository: https://github.com/moucheng2017/EMSSL.

TdDS-UNet: top-down deeply supervised U-Net for the delineation of 3D colorectal cancer

Automatically delineating colorectal cancers with fuzzy boundaries from 3D images is a challenging task, but the problem of fuzzy boundary delineation in existing deep learning-based methods have not been investigated in depth. Here, an encoder–decoder-based U-shaped network (U-Net) based on top-down deep supervision (TdDS) was designed to accurately and automatically delineate the fuzzy boundaries of colorectal cancer. TdDS refines the semantic targets of the upper and lower stages by mapping ground truths that are more consistent with the stage properties than upsampling deep supervision. This stage-specific approach can guide the model to learn a coarse-to-fine delineation process and improve the delineation accuracy of fuzzy boundaries by gradually shrinking the boundaries. Experimental results showed that TdDS is more customizable and plays a role similar to the attentional mechanism, and it can further improve the capability of the model to delineate colorectal cancer contours. A total of 103, 12, and 29 3D pelvic magnetic resonance imaging volumes were used for training, validation, and testing, respectively. The comparative results indicate that the proposed method exhibits the best comprehensive performance, with a dice similarity coefficient (DSC) of 0.805 ± 0.053 and a hausdorff distance (HD) of 9.28 ± 5.14 voxels. In the delineation performance analysis section also showed that 44.49% of the delineation results are satisfactory and do not require revisions. This study can provide new technical support for the delineation of 3D colorectal cancer. Our method is open source, and the code is available at https://github.com/odindis/TdDS/tree/main.

Is vestibular function related to human hippocampal volume?

Recent studies implicate the effect of vestibular loss on cognitive decline, including hippocampal volume loss. As hippocampal atrophy is an important biomarker of Alzheimer's disease, exploring vestibular dysfunction as a risk factor for dementia and its role in hippocampal atrophy is of interest. To replicate previous literature on whole-brain and hippocampal volume in semicircular canal dysfunction (bilateral vestibulopathy; BV) and explore the association between otolith function and hippocampal volume. Hippocampal and whole-brain MRI volumes were compared in adults aged between 55 and 83 years. Participants with BV (n = 16) were compared to controls individually matched on age, sex, and hearing status (n = 16). Otolith influence on hippocampal volume in preserved semicircular canal function was evaluated (n = 34). Whole-brain and targeted hippocampal approaches using volumetric and surface-based measures yielded no significant differences when comparing BV to controls. Binary support vector machines were unable to classify inner ear health status above chance level. Otolith parameters were not associated with hippocampal volume in preserved semicircular canal function. No significant differences in whole-brain or hippocampal volume were found when comparing BV participants with healthy controls. Saccular parameters in subjects with preserved semicircular canal function were not associated with hippocampal volume changes.

OMERACT validation of a deep learning algorithm for automated absolute quantification of knee joint effusion versus manual semi-quantitative assessment

ObjectiveTo begin evaluating deep learning (DL)-automated quantification of knee joint effusion-synovitis via the OMERACT filter. MethodsA DL algorithm previously trained on Osteoarthritis Initiative (OAI) knee MRI automatically quantified effusion volume in MRI of 53 OAI subjects, which were also scored semi-quantitatively via KIMRISS and MOAKS by 2–6 readers. ResultsDL-measured knee effusion correlated significantly with experts’ assessments (Kendall's tau 0.34–0.43) ConclusionThe close correlation of automated DL knee joint effusion quantification to KIMRISS manual semi-quantitative scoring demonstrated its criterion validity. Further assessments of discrimination and truth vs. clinical outcomes are still needed to fully satisfy OMERACT filter requirements.

Explainable hybrid vision transformers and convolutional network for multimodal glioma segmentation in brain MRI

Accurate localization of gliomas, the most common malignant primary brain cancer, and its different sub-region from multimodal magnetic resonance imaging (MRI) volumes are highly important for interventional procedures. Recently, deep learning models have been applied widely to assist automatic lesion segmentation tasks for neurosurgical interventions. However, these models are often complex and represented as “black box” models which limit their applicability in clinical practice. This article introduces new hybrid vision Transformers and convolutional neural networks for accurate and robust glioma segmentation in Brain MRI scans. Our proposed method, TransXAI, provides surgeon-understandable heatmaps to make the neural networks transparent. TransXAI employs a post-hoc explanation technique that provides visual interpretation after the brain tumor localization is made without any network architecture modifications or accuracy tradeoffs. Our experimental findings showed that TransXAI achieves competitive performance in extracting both local and global contexts in addition to generating explainable saliency maps to help understand the prediction of the deep network. Further, visualization maps are obtained to realize the flow of information in the internal layers of the encoder-decoder network and understand the contribution of MRI modalities in the final prediction. The explainability process could provide medical professionals with additional information about the tumor segmentation results and therefore aid in understanding how the deep learning model is capable of processing MRI data successfully. Thus, it enables the physicians’ trust in such deep learning systems towards applying them clinically. To facilitate TransXAI model development and results reproducibility, we will share the source code and the pre-trained models after acceptance at https://github.com/razeineldin/TransXAI.

Knee cartilage MR images segmentation based on multi-dimensional hybrid convolutional neural network

Accurate segmentation of knee cartilage in MR images is crucial for early diagnosis and treatment of knee conditions. Manual segmentation is time-consuming, leading researchers to explore automatic deep learning methods. However, the choice between 2D and 3D networks for organ segmentation remains debated. In this paper, we propose a hybrid 2D and 3D deep neural network approach, named UVNet, which combines the strengths of both techniques to enhance segmentation performance. Within this network structure, the 3D segmentation network serves as the backbone for feature extraction, while the 2D segmentation network functions as an information supplement network. Local and global MIP images are generated by employing various maximum intensity projection modes of knee MRI volumes as input for the information supplement network. By constructing a local and global MIP feature fusion module, the supplementary information obtained from the 2D segmentation network is fully integrated into the backbone network. We assess the quality of the proposed method using the Osteoarthritis Initiative (OAI) dataset and the 2010 Grand Challenge Knee Image Segmentation (SKI-10) dataset, comparing it to the Baseline Network and other advanced 2D and 3D segmentation methods. The experiments demonstrate that UVNet achieves competitive performance in the aforementioned two cartilage segmentation tasks.

Precise and Rapid Whole-Head Segmentation from Magnetic Resonance Images of Older Adults using Deep Learning.

Whole-head segmentation from Magnetic Resonance Images (MRI) establishes the foundation for individualized computational models using finite element method (FEM). This foundation paves the path for computer-aided solutions in fields, particularly in non-invasive brain stimulation. Most current automatic head segmentation tools are developed using healthy young adults. Thus, they may neglect the older population that is more prone to age-related structural decline such as brain atrophy. In this work, we present a new deep learning method called GRACE, which stands for General, Rapid, And Comprehensive whole-hEad tissue segmentation. GRACE is trained and validated on a novel dataset that consists of 177 manually corrected MR-derived reference segmentations that have undergone meticulous manual review. Each T1-weighted MRI volume is segmented into 11 tissue types, including white matter, grey matter, eyes, cerebrospinal fluid, air, blood vessel, cancellous bone, cortical bone, skin, fat, and muscle. To the best of our knowledge, this work contains the largest manually corrected dataset to date in terms of number of MRIs and segmented tissues. GRACE outperforms five freely available software tools and a traditional 3D U-Net on a five-tissue segmentation task. On this task, GRACE achieves an average Hausdorff Distance of 0.21, which exceeds the runner-up at an average Hausdorff Distance of 0.36. GRACE can segment a whole-head MRI in about 3 seconds, while the fastest software tool takes about 3 minutes. In summary, GRACE segments a spectrum of tissue types from older adults T1-MRI scans at favorable accuracy and speed. The trained GRACE model is optimized on older adult heads to enable high-precision modeling in age-related brain disorders. To support open science, the GRACE code and trained weights are made available online and open to the research community at https://github.com/lab-smile/GRACE.