MRI Images Research Articles

Desmoid-type fibromatosis of the breast in a male patient following cosmetic surgery: a rare case report

BackgroundDesmoid-type fibromatosis of the breast is a rare, benign, but locally aggressive tumor that typically affects women. Its presentation in male patients is exceedingly rare, and even more so following a cosmetic procedure such as liposuction. This case report describes a unique presentation of breast fibromatosis in a male patient, who developed the condition after undergoing liposuction for cosmetic purposes to define the pectoral area. The case highlights the diagnostic challenges it poses due to its malignancy-like appearance on imaging.Materials and methodsA 28-year-old male patient presented with a right palpable mass in the breast following a liposuction procedure aimed at enhancing pectoral definition. Imaging studies, including ultrasound and MRI, raised suspicion of a malignancy. However, histopathological analysis from a core needle biopsy revealed fibromatosis. The patient underwent surgical excision of the tumor, and histological evaluation confirmed the diagnosis of benign desmoid-type breast fibromatosis.ResultsHistopathological examination revealed benign fibromatosis with no evidence of malignancy. The tumor was completely excised with clear margins, and the patient has had no signs of recurrence during the follow-up period.ConclusionsThis case highlights the rare occurrence of breast fibromatosis in a male patient following liposuction for cosmetic purposes. Given its ability to mimic malignancy on imaging, early diagnosis and complete surgical excision are essential for effective management and to prevent recurrence.

Locus coeruleus signal intensity and emotion regulation in agitation in Alzheimer’s disease

Abstract Hyperphosphorylated tau accumulation is seen in the noradrenergic locus coeruleus from the earliest stages of Alzheimer’s disease onwards and has been associated with symptoms of agitation. It is hypothesized that compensatory locus coeruleus-noradrenaline system overactivity and impaired emotion regulation could underlie agitation propensity, but to our knowledge this has not previously been investigated. A better understanding of the neurobiological underpinnings of agitation would help the development of targeted prevention and treatment strategies. Using a sample of individuals with amnestic mild cognitive impairment and probable mild Alzheimer’s disease dementia from the German Center for Neurodegenerative Diseases (DZNE)-Longitudinal Cognitive Impairment and Dementia (DELCODE) study cohort (N=309, aged 67-96 years, 51% female), we assessed cross-sectional relationships between a latent factor representing the functional integrity of an affect-related executive regulation network and agitation point prevalence and severity scores. In a subsample of individuals with locus coeruleus MRI imaging data (N=37, aged 68-93 years, 49% female), we also investigated preliminary associations between locus coeruleus MRI contrast ratios (a measure of structural integrity, whole or divided into rostral, middle, and caudal thirds) and individual affect-related regulation network factor scores and agitation measures. Regression models controlled for effects of age and clinical disease severity and, for models including resting state functional MRI connectivity variables, grey matter volume and education years. Agitation point prevalence showed a positive relationship with a latent factor representing the functional integrity (and a negative relationship with a corresponding structural measure) of the affect-related executive regulation network. Locus coeruleus MRI contrast ratios were positively associated with agitation severity (but only for the rostral third, in N=13) and negatively associated with the functional affect-related executive regulation latent factor scores. Resting-state functional connectivity between a medial prefrontal cortex region and left amygdala was related to locus coeruleus MRI contrast ratios. These findings implicate the involvement of locus coeruleus integrity and emotion dysregulation in agitation in Alzheimer’s disease and support the presence of potential compensatory processes. At the neural level, there may be a dissociation between mechanisms underlying agitation risk per se and symptom severity. Further studies are needed to replicate and extend these findings, incorporating longitudinal designs, measures of autonomic function and non-linear modeling approaches to explore potential causal and context-dependent relationships across Alzheimer’s disease stages.

Differentiation of histological calcification classifications in breast cancer using ultrashort echo time and chemical shift-encoded imaging MRI

IntroductionDuctal carcinoma in situ (DCIS) accounts for 25% of newly diagnosed breast cancer cases with only 14%–53% developing into invasive ductal carcinoma (IDC), but currently overtreated due to inadequate accuracy of mammography. Subtypes of calcification, discernible from histology, has been suggested to have prognostic value in DCIS, while the lipid composition of saturated and unsaturated fatty acids may be altered in de novo synthesis with potential sensitivity to the difference between DCIS and IDC. We therefore set out to examine calcification using ultra short echo time (UTE) MRI and lipid composition using chemical shift-encoded imaging (CSEI), as markers for histological calcification classification, in the initial ex vivo step towards in vivo application.MethodsTwenty female patients, with mean age (range) of 57 (35–78) years, participated in the study. Intra- and peri-tumoural degree of calcification and peri-tumoural lipid composition were acquired on MRI using UTE and CSEI, respectively. Ex vivo imaging was conducted on the freshly excised breast tumour specimens immediately after surgery. Histopathological analysis was conducted to determine the calcification status, Nottingham Prognostic Index (NPI), and proliferative activity marker Ki-67.ResultsIntra-tumoural degree of calcification in malignant classification (1.05 ± 0.13) was significantly higher (p = 0.012) against no calcification classification (0.84 ± 0.09). Peri-tumoural degree of calcification in malignant classification (1.64 ± 0.10) was significantly higher (p = 0.033) against no calcification classification (1.41 ± 0.18). Peri-tumoural MUFA in malignant classification (0.40 ± 0.01) was significantly higher (p = 0.039) against no calcification classification (0.38 ± 0.02). Ki-67 showed significant negative correlation against peri-tumoural MUFA (p = 0.043, ρ = −0.457), significant positive correlation against SFA (p = 0.008, ρ = 0.577), and significant negative correlation against PUFA (p = 0.002, ρ = −0.653).ConclusionThe intra- and peri-tumoural degree of calcification and peri-tumoural MUFA are sensitive to histological calcification classes supporting future investigation into DCIS prognosis.

Anogenital distance on MRI does not correlate to surgical diagnosis of endometriosis in patients without prior abdominal surgery

Anogenital distance (AGD) is regarded as a potential biomarker for endometriosis, and a measurement on MRI images has been found to be promising. This study aimed to evaluate the measurement of AGD on MRI to predict the surgical diagnosis of endometriosis. We included 127 patients who received an MRI for endometriosis between October 2018 and February 2023. AGD was measured on MRI by two readers (MRI-AGD-AC: clitoris to anus; MRI-AGD-AF: posterior fourchette to anus). The feasibility and interobserver reliability of AGD measurements were evaluated. Differences in AGD between patient groups were analyzed. The intraclass correlation coefficient estimates indicated a good to excellent reliability of MRI-AGD-AC (0.92; 95% CI: 0.83–0.95) and a poor to good reliability of MRI-AGD-AF (0.68; 95% CI: 0.27–0.83). No statistically significant differences in the mean MRI-AGD-AC and MRI-AGD-AF in patients with and without surgical diagnosis of DIE (p = 0.413; p = 0.110), peritoneal endometriosis with and without DIE (p = 0.641; p = 0.323), and ovarian endometriosis (p = 0.155; p = 0.150) were found. The AUC ranged from 0.475 (95% CI: 0.365–0.584) to 0.586 (95% CI: 0.454–0.718). Thus, AGD does not constitute a valuable biomarker for patients with clinically suspected endometriosis.

Approximating R1 and R2: A Quantitative Approach to Clinical Weighted MRI.

Weighted MRI images are widely used in clinical as well as open-source neuroimaging databases. Weighted images such as T1-weighted, T2-weighted, and proton density-weighted (T1w, T2w, and PDw, respectively) are used for evaluating the brain's macrostructure; however, their values cannot be used for microstructural analysis, as they lack physical meaning. Quantitative MRI (qMRI) relaxation rate parameters (e.g., R1 and R2) do contain microstructural physical meaning. Nevertheless, qMRI is rarely done in large-scale clinical databases. Currently, the weighted images ratio T1w/T2w is used as a quantifier to approximate the brain's microstructure. In this paper, we test three additional quantifiers that approximate quantitative maps, which can help bring quantitative MRI to the clinic for easy use. Following the signal equations and using simple mathematical operations, we combine the T1w, T2w, and PDw images to estimate the R1 and R2 maps. We find that two of these quantifiers (T1w/PDw and T1w/ln(T2w)) can approximate R1, and that (ln(T2w/PDw)) can approximate R2, in 3 datasets that were tested. We find that this approach also can be applied to T2w scans taken from widely available DTI (Diffusion Tensor Imaging)datasets. We tested these quantifiers on both invitro phantom and invivo human datasets. We found that the quantifiers accurately represent the quantitative parameters across datasets. Finally, we tested the quantifiers within a clinical context, and found that they are robust across datasets. Our work provides a simple pipeline to enhance the usability and quantitative accuracy of MRI weighted images.

3D Non-separable Moment Invariants and Their Use in Neural Networks

Recognition of 3D objects is an important task in many bio-medical and industrial applications. The recognition algorithms should work regardless of a particular orientation of the object in the space. In this paper, we introduce new 3D rotation moment invariants, which are composed of non-separable Appell moments. We show that non-separable moments may outperform the separable ones in terms of recognition power and robustness thanks to a better distribution of their zero surfaces over the image space. We test the numerical properties and discrimination power of the proposed invariants on three real datasets—MRI images of human brain, 3D scans of statues, and confocal microscope images of worms. We show the robustness to resampling errors improved more than twice and the recognition rate increased by 2–10 % comparing to most common descriptors. In the last section, we show how these invariants can be used in state-of-the-art neural networks for image recognition. The proposed H-NeXtA architecture improved the recognition rate by 2–5 % over the current networks.

Diagnosis of Acute Versus Chronic Thoracolumbar Vertebral Compression Fractures Using CT Radiomics Based on Machine Learning: a Preliminary Study.

The purpose of this study is to evaluate the performance of radiomic models in acute thoracolumbar vertebral compression fractures (VCFs) and their impact on radiologists. In this monocentre retrospective study, eligible for inclusion were adults who underwent emergent thoracic/lumbar CT between May 2022 and November 2023 in our hospital diagnosed with thoracolumbar VCFs. The lesions were randomly divided at a ratio of 7:3 into a training set and test set. For external validation, consecutive patients who underwent emergent thoracic/lumbar CT between January 2022 and April 2022 were included. MRI and previous imaging were used as reference standard. The vertebral body area was manually segmented. Logistic regression was used to construct a CT radiomic model and a combined model, including Relief-selected radiomic features and clinical information. The radiologists' diagnosis with and without the models was recorded. The performance was assessed using receiver operating characteristic curves (ROC), calibration curves (CC) and decision curve analysis (DCA). Of 235 VCFs in 147 patients (median age, 73years, 66 male) included, the diagnosis of acute VCFs was confirmed in 126. The area under the ROC of the CT radiomics model and the combined model in the external validation set were 0.883 (95% CI 0.777, 0.998) and 0.875 (95% CI 0.768, 0.982), respectively. CC and DCA showed good clinical application of the models. The less experienced reader achieved a higher accuracy with the help of the models (p = 0.027). The radiomic models showed high accuracy for diagnosing acute VCFs and helped radiologists improve the accuracy of diagnosis.

Robustness of 4D flow MRI derived aortic wall shear stress and pulse wave velocity across different protocols in healthy controls and in patients with bicuspid aortic valve.

To evaluate the reproducibility of important biomarkers like wall shear stress (WSS), pulse wave velocity (PWV), and net flow across two 4D flow MRI imaging protocols with different coverages: aorta-targeted 4D flow MRI (AT4D) and whole-heart 4D flow (WH4D) protocols. Thirty-eight control subjects (43.2 ± 10.1 years old; 22 males) and ten patients (45.7 ± 8.9 years old; 7 males) with bicuspid aortic valve (BAV) were included. Each subject underwent AT4D and WH4D scans. Absolute WSS, PWV, and net flow were assessed for each patient across both protocols and compared using Bland-Altman analysis. Areas of elevated WSS were assessed for BAV patients across different WSS thresholds that define WSS to be elevated compared to a normal population average. A sensitivity analysis was conducted to determine the best WSS threshold at which WH4D-derived areas most closely resemble AT4D-derived areas. Inter-rater reproducibility was evaluated in twenty-four subjects. AT4D and WH4D PWV and WSS estimates demonstrated good agreement (PWV: -0.12 ± 1.84m/s, p = 0.4; Median WSS: 0.06 ± 0.13Pa, p < 0.01; Maximum WSS: 0.04 ± 0.27Pa, p = 0.07). Good agreement was also found for AAo net flow (8.14 ± 24.86 mL/cycle, p < 0.01). PWV correlated with age across protocols (AT4D: r = 0.68, p < 0.01; WH4D: r = 0.72, p < 0.01). Sensitivity analysis identified a WSS threshold where WH4D-derived areas of elevated WSS most closely resembled AT4D-derived areas. Inter-rater assessment of the tested parameters resulted in a small mean difference percentage of < 3%. PWV, WSS, and net flow demonstrated good agreement across protocols. The WSS threshold should be adjusted for WH4D estimates to optimally match AT4D-derived output. Reproducibility analysis showed good test-retest agreement. This study demonstrates the reproducibility of certain hemodynamic parameters across two 4D flow MRI protocol.

A Novel GAN-based Chaotic Method with DNA Computing for Enhancing Security of Medical Images

Medical images are utilized to diagnose patients' health conditions. Nowadays, medical images are sent over the internet for diagnosis purposes. So, they should be protected from cyber attackers. These medical images are sensitive to any minor changes, and the data volume is rapidly increasing. Thus, security and storage costs must be considered in medical images. Traditional encryption and compression methods are ineffective for encrypting medical images due to their high execution time and algorithm complexity. In this paper, a novel 2D-chaos and Generative Adversarial Network (GAN) with DeoxyriboNucleic Acid (DNA) computing is proposed for generating encryption keys and improving the security of medical images. The proposed scheme uses GAN and 2D-chaos to generate the private key and diffusion process. The pixel values of the original images in the proposed schemes are shuffled using Mersenne Twister (MT) to improve the security of medical images. Moreover, the novel 2D-Chaotic Tent Map (2D-CTM) method is used to construct the key while performing XOR-based encryption. The proposed model has been tested on different medical images, namely the COVIDx-19 X-ray images, the malaria microscopic images, and the brain MRI images. The experiment results have been evaluated using performance metrics, namely key space, histogram analysis, entropy, key sensitivity, robustness analysis, correlation, SSIM, and MS-SSIM. The outcomes demonstrate that the proposed scheme is more effective than the state-of-the-art schemes.

Radiation-induced nasopharyngeal necrosis combined with local recurrence in nasopharyngeal carcinoma: diagnosis and treatment strategies

ABSTRACT Background To identify the diagnosis and treatment strategies by analyzing the clinical characteristics and treatment methods of RNNCLR. Methods A total of 210 patients pathologically diagnosed with RNNCLR were retrospectively included. Clinical characteristics, MRI features, treatment methods, and survival outcomes were analyzed. Propensity score matching (PSM) analysis was performed to adjust the surgical benefit. Results Ninety-one patients (43.3%) took a single biopsy, 67 patients (31.9%) underwent repeated biopsies and 52 patients (24.8%) received endoscopic surgery to obtain pathological positive tissues. RNNCLR had characteristic imaging features distinguished from pure radiation necrosis. The interval from the previous radiotherapy was 13.2 (7.0, 23.3) months. The 1-year, 3-year, and 5-year overall survival rates were 59.6%, 32.3%, and 14.6%. Patients with reirradiation, detectable EBV-DNA level, or ICA exposure had a worse prognosis. Overall survival was significantly higher in the endoscopic surgery group than in nonsurgery group after PSM (3-year OS rates, 44.5% vs. 23.9%, p = 0.011). Conclusions Histopathological diagnosis of RNNCLR needs repeated biopsies or even surgery. Careful analysis of MRI images, correlation with interval time from last radiation, and short-term follow-up may solve the diagnostic dilemmas. Endoscopic surgery results in a survival benefit by completely resecting lesions or removing necrotic tissue to reduce necrosis-related complications.

PLB2Tau mice are impaired in novel and temporal object recognition and show corresponding traits in brain MRI

Recent clinical trials targeting tau protein aggregation have heightened interest in tau-based therapies for dementia. Success of such treatments depends crucially on translation from non-clinical animal models. Here, we present the age profile of the PLB2Tau knock-in model of fronto-temporal dementia in terms of cognition, and by utilising a directly translatable magnetic resonance imaging approach. Separate cohorts of mice aged 3, 6 and 12 months were tested in an object recognition protocol interrogating visual, spatial, and temporal discrimination in consecutive tests. Upon completion of their behavioural testing, animals were recorded in a 7T MRI for brain structural integrity and diffusion tensor imaging (DTI) analysis.We report that PLB2Tau mice presented with an age-dependent deficit in novel object discrimination relative to wild-type controls at 6 and 12 months. Spatial and temporal discrimination, though not significantly different from controls, appeared extremely challenging for PLB2Tau subjects, especially at 12 months, since they explored objects less than controls and were devoid of memory. Controls readily recalled all relevant object-related information. At the same time, the T2 weighted voxel-based image analysis revealed a progressive shrinkage of total brain volumes in 6- and 12-month-old PLB2Tau mice as well as relative striatal, but not hippocampal volumes. A regional DTI analysis yielded only reduced mean diffusivity of the fimbria, but not CA1 or dentate gyrus, amygdala, cingulate cortex, or corpus callosum.These data confirm the PLB2Tau mouse as a translationally useful model for dementia research and suggest the importance of the hippocampal input as a determinant for novel object discrimination.

Quantitative assessment of cervical disc degeneration using disc signal intensity index.

The assessment of disc degeneration remains a significant challenge in clinical research. Pfirrmann grade is a frequently used classification for lumbar disc degeneration on MRI. However, there has been no gold standard for cervical spine disc degeneration. Recently, we introduced the Disc Signal Intensity Index (DSI2) as a quantitative disc assessment for the lumbar spine, which is easily measurable in the cervical spine. The aim of this study was to apply DSI2 in the cervical intervertebral disc and investigate the factors associated with the cervical disc degeneration. Cross-sectional study using retrospectively collected data. Cervical MRIs from a database of patients undergoing ACDF between 2015 and 2018 were retrospectively reviewed. Demographic variables included age, sex, body mass index (BMI), race, smoking status, and comorbidities such as diabetes, chronic kidney disease, and coronary artery disease. DSI2 measurements were performed on midsagittal T2-weighted MRI images by determining the intensity within regions of interest (ROI). One ROI was set in the cerebrospinal fluid (CSF) and three ROIs were set per disc at the anterior, middle, and posterior third. The mean of the three measurements per disc was then divided by that of the CSF to calculate the DSI2 score. Multivariable linear regression analyses with mixed model were conducted to determine the potential contributing factors for disc degeneration. A total of 149 patients and 770 discs were included in the final analysis. Ninety-three patients (37.6%) were female and the mean (SD) age was 55.6 (11.7) years. The distribution of DSI2 scores among the different Pfirrmann grades was as follows: Grade 1: 0.259±NA; Grade 2: 0.226±0.090; Grade 3: 0.175±0.070; Grade 4: 0.136±0.060; Grade 5: 0.131±0.050. Multivariable linear mixed-effect regression analysis, setting with DSI2 as the objective variable, demonstrated that age (β= -0.130, p< 0.05), BMI (β= -2.06, p< 0.05), Modic changes (Type1 β= -2.70, p<0.01) were independent contributors to disc degeneration. The segments C4/5 and C7/T1 were less prone to disc degeneration (C4/5: β=1.37, p<0.001; C7/T1: β= 2.63, p<0.001) and the history of diabetes (β=5.31, p<0.01) was associated with high DSI2.(p<0.01) CONCLUSIONS: The present study provides valuable insights for identifying risk factors in degenerative cervical conditions utilizing the DSI2. The DSI2 method emerges as a promising alternative for future disc research, excelling in the detection of subtle progressions of degeneration and distinguishing itself from the subjective Pfirrmann grading system.

Assessing the Impact of Cardiac Output and Valve Orientation on Bioprosthetic Pulmonary Valve Hemodynamics Using In Vitro 4D-Flow MRI and High-Speed Imaging.

Pulmonary valve replacement (PVR) using bioprosthetic valves is a common procedure performed in patients with repaired Tetralogy of Fallot and other conditions, but these valves frequently become dysfunctional within 15years of implantation. The causes for early valve failure are not clearly understood. The purpose of this study was to explore the impact of changing cardiac output (CO) and valve orientation on local hemodynamics and valve performance. A 25mm bioprosthetic valve was implanted in an idealized 3D-printed model of the right ventricular outflow tract (RVOT). The local hemodynamics at three COs and two valve orientations were assessed using 4D-Flow MRI and high-speed camera imaging. Noticeable differences in jet asymmetry, the amount of recirculation, leaflet opening patterns, as well as the size and location of reversed flow regions were observed with varying CO. Rotation of the valve resulted in drastic differences in reversed flow regions, but not forward flow. Flow features observed in the valve with low CO in this study have previously been correlated with calcification, hemolysis, and leaflet fatigue, indicating their potential negative impact on local hemodynamics and leaflet performance.

Beware of diffusion models for synthesizing medical images - A comparison with GANs in terms of memorizing brain MRI and chest x-ray images

Abstract Diffusion models were initially developed for text-to-image generation and are now being utilized to generate high quality synthetic images. Preceded by GANs, diffusion models have shown impressive results using various evaluation metrics. However, commonly used metrics such as FID and IS are not suitable for determining whether diffusion models are simply reproducing the training images. Here we train StyleGAN and a diffusion model, using BRATS20, BRATS21 and a chest x-ray pneumonia dataset, to synthesize brain MRI and chest x-ray images, and measure the correlation between the synthetic images and all training images. Our results show that diffusion models are more likely to memorize the training images, compared to StyleGAN, especially for small datasets and when using 2D slices from 3D volumes. Researchers should be careful when using diffusion models (and to some extent GANs) for medical imaging, if the final goal is to share the synthetic images.

DenseIncepS115: a novel network-level fusion framework for Alzheimer's disease prediction using MRI images.

One of the most prevalent disorders relating to neurodegenerative conditions and dementia is Alzheimer's disease (AD). In the age group 65 and older, the prevalence of Alzheimer's disease is increasing. Before symptoms showed up, the disease had grown to a severe stage and resulted in an irreversible brain disorder that is not treatable with medication or other therapies. Therefore, early prediction is essential to slow down AD progression. Computer-aided diagnosis systems can be used as a second opinion by radiologists in their clinics to predict AD using MRI scans. In this work, we proposed a novel deep learning architecture named DenseIncepS115for for AD prediction from MRI scans. The proposed architecture is based on the Inception Module with Self-Attention (InceptionSA) and the Dense Module with Self-Attention (DenseSA). Both modules are fused at the network level using a depth concatenation layer. The proposed architecture hyperparameters are initialized using Bayesian Optimization, which impacts the better learning of the selected datasets. In the testing phase, features are extracted from the depth concatenation layer, which is further optimized using the Catch Fish Optimization (CFO) algorithm and passed to shallow wide neural network classifiers for the final prediction. In addition, the proposed DenseIncepS115 architecture is interpreted through Lime and Gradcam explainable techniques. Two publicly available datasets were employed in the experimental process: Alzheimer's ADNI and Alzheimer's classes MRI. On both datasets, the proposed architecture obtained an accuracy level of 99.5% and 98.5%, respectively. Detailed ablation studies and comparisons with state-of-the-art techniques show that the proposed architecture outperforms.

Prediction of Brain Cancer Occurrence and Risk Assessment of Brain Hemorrhage Using Hybrid Deep Learning Technique

The prediction of brain cancer occurrence and risk assessment of brain hemorrhage using a hybrid deep learning (DL) technique is a critical area of research in medical imaging analysis. One prominent challenge in this field is the accurate identification and classification of brain tumors and hemorrhages, which can significantly impact patient prognosis and treatment planning. The objectives of the study address the prediction of brain cancer occurrence and the assessment of risk levels associated with both brain cancers due to brain hemorrhage. A diverse dataset of brain MRI and CT scan images. Utilize Unsymmetrical Trimmed Median Filter with Optics Clustering for noise removal while preserving edges and details. The Chan-Vese segmentation process for refined segmentation. Brain cancer detection using Multi-Head Self-Attention Dilated Convolution Neural Network (MH-SA-DCNN) with Efficient Net Model. Brain cancer detection using MH-SA-DCNN with Efficient Net Model. This trains the algorithm to predict cancerous regions in brain images. Further, implement a Graph-Based Deep Neural Network Model (G-DNN) to capture spatial relationships and risk factors from brain images. Cox regression model to estimate cancer risk over time and fine-tune and optimize the model’s parameters and features using the Osprey optimization algorithm (OPA).

Human-AI collaborative multi-modal multi-rater learning for endometriosis diagnosis.

Endometriosis, affecting about 10% of individuals assigned female at birth, is challenging to diagnose and manage. Diagnosis typically involves the identification of various signs of the disease using either laparoscopic surgery or the analysis of T1/T2 MRI images, with the latter being quicker and cheaper but less accurate. A key diagnostic sign of endometriosis is the obliteration of the Pouch of Douglas (POD). However, even experienced clinicians struggle with accurately classifying POD obliteration from MRI images, which complicates the training of reliable AI models. In this paper, we introduce the Human-AI Collaborative Multi-modal Multi-rater Learning (HAICOMM) methodology to address the challenge above. HAICOMM is the first method that explores three important aspects of this problem: 1) multi-rater learning to extract a cleaner label from the multiple "noisy" labels available per training sample; 2) multi-modal learning to leverage the presence of T1/T2 MRI images for training and testing; and 3) human-AI collaboration to build a system that leverages the predictions from clinicians and the AI model to provide more accurate classification than standalone clinicians and AI models. Presenting results on the multi-rater T1/T2 MRI endometriosis dataset that we collected to validate our methodology, the proposed HAICOMM model outperforms an ensemble of clinicians, noisy-label learning models, and multi-rater learning methods.

The hypointense pulvinar sign on susceptibility weighed magnetic resonance imaging: A visual biomarker for iron deposition in epilepsy

Objective Our study aimed to investigate potential alterations in iron deposition within pulvinar, using susceptibility weighted imaging (SWI) MRI in epilepsy patients through a biomarker termed the “hypointense pulvinar sign.” Methods A full-text radiological information system search of radiological reports was performed for the term “epilepsy” between 2014 and 2022. Only patients with the diagnosis of epilepsy were included. SWI was assessed by two readers recording lateralization of an asymmetrically more hypointense pulvinar. Cohen’s kappa for inter-rater reliability was calculated. Fisher’s exact test was performed to assess for significance between groups. Results Our epilepsy cohort comprised 105 patients with following diagnoses: 45 intra-axial tumor, 13 meningioma, 13 MRI negative, 12 encephalomalacia, seven siderosis, six cavernoma, five arteriovenous malformation, two acute demyelinating encephalomyelitis, one tuberous sclerosis, one giant aneurysm. The hypointense pulvinar sign was correct in 44% of cases. Notably, right hemispheric lesions exhibited a significantly higher proportion of correct hypointense pulvinar signs compared to the left hemisphere (46% vs 24%; p = 0.044). Inter-rater reliability was substantial at 0.62 ( p &lt; 0.001). Only two of 21 (10%) of healthy controls demonstrated a hypointense pulvinar sign, which was significantly different from the epilepsy cohort ( p &lt; 0.01). Conclusions The hypointense pulvinar sign has proven to be a reproducible, simple to use biomarker for iron deposition in epilepsy which could be considered for inclusion into multimodal precision medicine models.

A dual phase biologic augmentation of rotator cuff healing in a preclinical rat model using IL-17F and low-dose lithium

BackgroundRotator cuff tendon tears are a common cause of shoulder dysfunction in adults. Yet, impaired healing continues to result in higher failures rates after surgical repair resulting in patient dysfunction and prolonged recovery. This has spurred increased investigation of biologic augmentation to improve tendon healing. This study examines the outcome of peritendinous IL-17F administration and oral low-dose lithium carbonate on rotator cuff healing following acute surgical repair in a rat model. MethodsTreatment group included an open supraspinatus peel and repair followed by a local injection of IL-17F at the bone-tendon interface and a 7-day course of oral lithium carbonate. Control rats received no additional intervention prior to surgical closure. Evaluation of healing was then preformed using MRI imaging, biomechanical testing, and histological analysis at the bone-tendon interface. ResultsEighteen rats (9 control, 9 experimental) underwent complete testing. Combined treatment of peri-tendinous IL-17F and oral low-dose lithium carbonate after rotator cuff repair improved rotator cuff healing in all outcomes when compared to controls. MRI imaging in the treatment group showed complete healing of all supraspinatus tendons across the anatomic footprint after repair. We also found significant increases in biomechanical stiffness compared to controls (p <0.01). At the histological level, treatment groups also had decreased osteoclasts (p<0.001), and increased histologic organization of fibroblasts compared to controls. These findings are consistent with an increase in biomechanical stiffness. ConclusionWe demonstrated that the combined treatment of IL-17F and oral low-dose lithium improved rotator cuff tendon healing quality following acute surgical repair in a rat model.

Bone Regrowth After Frontal Burr Hole Craniostomy: Natural History of 14-mm and 20-mm Burr Holes and Implications for Postoperative Trans-Burr Hole Ultrasound

BACKGROUND AND OBJECTIVE: Burr hole craniostomy is performed for ventriculoperitoneal shunt insertion and endoscopic third ventriculostomy in patients with cerebrospinal fluid disorders. These burr holes are increasingly being used as windows for postoperative ultrasound, an investigational alternative to computed tomography or MRI for follow-up imaging of ventricular caliber. However, bone regrowth reduces ultrasound visibility, and little is known about burr hole regrowth rates in adults. Our study evaluates burr hole regrowth patterns and implications for transcranial ultrasound imaging. METHODS: We retrospectively analyzed 101 consecutive patients who had frontal burr hole craniostomy for new ventriculoperitoneal shunt insertion or endoscopic third ventriculostomy over a 3-year period. A mix of standard 14-mm burr holes and expanded 20-mm burr holes were used. Burr hole bone regrowth was assessed using serial follow-up computed tomography scans. Linear and logistic regression analyses examined if bone regrowth correlated with any clinical variables. RESULTS: There was wide variability in rate and degree of burr hole regrowth. The average percentage closure was 25% at 6 months, with minimal additional closure over the following 18 months. The mean residual diameter for 14-mm and 20-mm burr holes stabilized around 9.4 mm and 15.4 mm, respectively. Bone regrowth was not associated with patient characteristics, including age, sex, skull thickness, or etiology of cerebrospinal fluid disorder. Rate of bone regrowth was similar between both cohorts. CONCLUSION: Bone regrowth after burr hole craniostomy is common, even in elderly patients, occurring rapidly within the first 6 to 12 months and subsequently stabilizing. It is frequently severe enough to restrict ultrasound visualization. Regrowth could not be predicted with any investigated variables, so uniform techniques are needed to block regrowth to allow for longitudinal ultrasound imaging, such as full-thickness cylindrical burr hole implants.