Tomography Magnetic Resonance Imaging Research Articles

A Three‐Step Automated Segmentation Method for Early Cervical Cancer MRI Images Based on Deep Learning

ABSTRACTTumor detection and segmentation are essential for cervical cancer (CC) treatment and diagnosis. This study presents a model that segmented the tumor, uterus, and vagina based on deep learning automatically on magnetic resonance imaging (MRI) images of patients with CC. The tumor detection dataset consists of 68 CC patients' diffusion‐weighted magnetic resonance imaging (DWI) images. The segmented dataset consists of 73 CC patients' T2‐weighted imaging (T2WI) images. First, the three clear images of the patient's DWI images are detected using a single‐shot multibox detector (SSD). Second, the serial number of the clearest image is obtained by scores, while the corresponding T2WI image with the same serial number is selected. Third, the selected images are segmented by employing the semantic segmentation (U‐Net) model with the squeeze‐and‐excitation (SE) block and attention gate (SE‐ATT‐Unet). Three segmentation models are implemented to automatically segment the tumor, uterus, and vagina separately by adding different attention mechanisms at different locations. The target detection accuracy of the model is 92.32%, and the selection accuracy is 90.9%. The dice similarity coefficient (DSC) on the tumor is 92.20%, pixel accuracy (PA) is 93.08%, and the mean Hausdorff distance (HD) is 3.41 mm. The DSC on the uterus is 93.63%, PA is 91.75%, and the mean HD is 9.79 mm. The DSC on the vagina is 75.70%, PA is 85.46%, and the mean HD is 10.52 mm. The results show that the proposed method accurately selects images for segmentation, and the SE‐ATT‐Unet is effective in segmenting different regions on MRI images.

The Classification of Metastatic Spine Cancer and Spinal Compression Fractures by Using CNN and SVM Techniques.

Metastatic spine cancer can cause pain and neurological issues, making it challenging to distinguish from spinal compression fractures using magnetic resonance imaging (MRI). To improve diagnostic accuracy, this study developed artificial intelligence (AI) models to differentiate between metastatic spine cancer and spinal compression fractures in MRI images. MRI data from Gyeongsang National University Hospital, collected from January 2019 to April 2022, were processed using Otsu's binarization and Canny edge detection algorithms. Using these preprocessed datasets, convolutional neural network (CNN) and support vector machine (SVM) models were built. The T1-weighted image-based CNN model demonstrated high sensitivity (1.00) and accuracy (0.98) in identifying metastatic spine cancer, particularly with data processed by Otsu's binarization and Canny edge detection, achieving exceptional performance in detecting cancerous cases. This approach highlights the potential of preprocessed MRI data for AI-assisted diagnosis, supporting clinical applications in distinguishing metastatic spine cancer from spinal compression fractures.

Postoperative Imaging of Bone and Soft Tissue Tumors in the Extremity: A Comprehensive Review.

Postoperative imaging of musculoskeletal tumors poses a significant diagnostic challenge for radiologists. The complexity arises from the need to differentiate between expected postoperative changes, potential complications, and local recurrence. The choice of imaging modality depends on the type of primary tumor. Standard radiological modalities such as radiography, computed tomography (CT), and magnetic resonance imaging (MRI) are widely utilized. Radiography and CT are especially valuable for assessing primary bone tumors, as they provide detailed views of bone structures and alignment, as well as revealing postoperative complications. MRIs are particularly effective for evaluating soft tissue tumors and identifying local recurrences due to its superior soft tissue contrast. The advanced imaging techniques, such as diffusion-weighted imaging and dynamic contrast-enhanced MRI, have significantly improved diagnostic accuracy in detecting tumor recurrence. An in-depth understanding of surgery-specific imaging findings, as well as the ability to detect recurrent disease, is crucial for early diagnosis of complications and improved patient outcomes. Familiarity with normal postoperative changes helps radiologists distinguish them from abnormal findings indicative of complications or tumor recurrence. This review article aims to outline the surgical options for musculoskeletal tumors, detail the various imaging techniques used in postoperative surveillance, and discuss the potential complications. By understanding the role of different imaging modalities and their applications associated with various surgical procedures, clinicians and radiologists can provide accurate and timely diagnoses.

Intramyocardial Hemangioma Mimicking Paraganglioma in Cardiac MRI and DOTANOC PET

AbstractA cardiac hemangioma is a rare primary benign intracardiac tumor characterized by proliferation of endothelial cells. The lesions are hyperintense on T2/short tau inversion recovery magnetic resonance imaging (MRI) and show avid enhancement and retention of contrast agents. Here we present a case of a histopathologically proven cardiac hemangioma showing uptake on Dodecanetetraacetic acid - octreotate (DOTATATE), masquerading as paraganglioma in MRI and positron emission tomography imaging.

ResNet-Vision Transformer based MRI-endoscopy fusion model for predicting treatment response to neoadjuvant chemoradiotherapy in locally advanced rectal cancer: A multicenter study.

Neoadjuvant chemoradiotherapy followed by radical surgery has been a common practice for patients with locally advanced rectal cancer, but the response rate varies among patients. This study aimed to develop a ResNet-Vision Transformer based magnetic resonance imaging (MRI)-endoscopy fusion model to precisely predict treatment response and provide personalized treatment. In this multicenter study, 366 eligible patients who had undergone neoadjuvant chemoradiotherapy followed by radical surgery at eight Chinese tertiary hospitals between January 2017 and June 2024 were recruited, with 2928 pretreatment colonic endoscopic images and 366 pelvic MRI images. An MRI-endoscopy fusion model was constructed based on the ResNet backbone and Transformer network using pretreatment MRI and endoscopic images. Treatment response was defined as good response or non-good response based on the tumor regression grade. The Delong test and the Hanley-McNeil test were utilized to compare prediction performance among different models and different subgroups, respectively. The predictive performance of the MRI-endoscopy fusion model was comprehensively validated in the test sets and was further compared to that of the single-modal MRI model and single-modal endoscopy model. The MRI-endoscopy fusion model demonstrated favorable prediction performance. In the internal validation set, the area under the curve (AUC) and accuracy were 0.852 (95% confidence interval [CI]: 0.744-0.940) and 0.737 (95% CI: 0.712-0.844), respectively. Moreover, the AUC and accuracy reached 0.769 (95% CI: 0.678-0.861) and 0.729 (95% CI: 0.628-0.821), respectively, in the external test set. In addition, the MRI-endoscopy fusion model outperformed the single-modal MRI model (AUC: 0.692 [95% CI: 0.609-0.783], accuracy: 0.659 [95% CI: 0.565-0.775]) and the single-modal endoscopy model (AUC: 0.720 [95% CI: 0.617-0.823], accuracy: 0.713 [95% CI: 0.612-0.809]) in the external test set. The MRI-endoscopy fusion model based on ResNet-Vision Transformer achieved favorable performance in predicting treatment response to neoadjuvant chemoradiotherapy and holds tremendous potential for enabling personalized treatment regimens for locally advanced rectal cancer patients.

Value of the Stretched Exponential and Fractional-order Model in Differentiating Hepatocellular from Intrahepatic Cholangiocarcinoma.

It remains unknown whether the parameters obtained using the Stretched Exponential Model (SEM) and Fractional Order Calculus (FROC) models can help distinguish Hepatocellular Carcinoma (HCC) from Intrahepatic Cholangiocarcinoma (ICC). This study aimed to evaluate the application value of the parameters of the 3.0T Magnetic Resonance Imaging (MRI) high-order SEM and FROC diffusion model in differentiating HCC and ICC. Patients with pathologically confirmed HCC and ICC were prospectively enrolled. Diffusion-weighted imaging scans with multiple b-values were acquired 2 weeks before the surgery. The original MRI images were fitted using the mono-exponential model, SEM, and FROC, and several parameters were obtained for the analysis. In total, 74 patients with HCC and 21 with ICC were included in the study. Significant differences between the HCC and ICC groups were noted in the Apparent Diffusion Coefficient (ADC: p = 0.007), Distributed Diffusion Coefficient (DDC: p < 0.001), and Diffusion coefficient (D: p < 0.001), as each value was significantly lower in the HCC than in the ICC group. The area under the receiver operating characteristic curve of ADC, DDC, and D was 0.694, 0.812, and 0.825, respectively, and the most effective corresponding cut-off values were 1.135 μm2/ms, 1.477 μm2/ms, and 1.104 μm2/ms, respectively. The diffusion parameters DDC from the SEM and D from the FROC model have been found to be more effective in discriminating HCC and ICC than the ADC from the mono-exponential model. Combining these quantitative parameters can improve the MRI's diagnostic accuracy, providing useful information for the preoperative differential diagnosis between HCC and ICC.

APPLICATION OF NEURAL NETWORKS IN SILICONE BREAST IMPLANT DIAGNOSTICS ON MAGNETIC RESONANCE IMAGING

Breast augmentation is one of the most frequently performed cosmetic procedures worldwide, but it carries certain risks including breast implant rupture. Timely and accurate diagnostics of ruptures are crucial, as undiagnosed ruptures can lead to serious health complications. Imaging methods, such as magnetic resonance imaging (MRI), are recommended for the diagnosis of breast implants due to their high accuracy. However, current diagnostics rely heavily on the subjective interpretation and experience of the physician. This study investigates the potential of neural networks (NN) to address this limitation and improve the accuracy of rupture detection in silicone breast implants. We applied a deep learning-based neural network system trained on MRI images of breast implants to detect ruptures. The dataset included annotated MRI scans of symptomatic and asymptomatic patients with confirmed implant integrity or rupture. Several models were trained using ResNet-18, ResNet-50, and Xception networks, with various hyperparameter settings and augmentation techniques applied to enhance model performance and generalizability. The performance of the NN model was evaluated using confusion matrices and standard metrics such as true positive rate (TPR) and true negative rate (TNR). A semi-automated algorithm for the detection of intracapsular ruptures of breast implants on MRI was successfully developed. The algorithm correctly detected ruptures in 95.4% of cases and accurately identified cases without rupture in 86.7% of instances. Our findings highlight the potential of neural networks as a supportive tool in diagnosing breast implant ruptures. By semi-automating rupture detection, NNs can reduce diagnostic errors, expedite image evaluation, and optimize resource use in medical practice. The study underscores the importance of combining artificial intelligence with expert evaluation to enhance patient care and reduce costs in medical diagnostics.

Labral size measured on preoperative magnetic resonance imaging not predictive of the need for labral reconstruction in patients undergoing primary hip arthroscopy

Abstract Preoperative radiographic measurements may help predict which patients with hip labral tears ultimately undergo repair versus primary reconstruction. This study investigated if radiographic parameters: (i) preoperatively predict labral repair versus reconstruction and (ii) correlate with T2 magnetic resonance imaging (MRI) mapping values of the labrum. This retrospective comparative study included patients aged 14–50 years who underwent labral repair or reconstruction at a single institution over a 2-year period. Patients with prior open or arthroscopic hip surgery or who had inadequate preoperative computed tomography (CT) and MRI imaging were excluded. Labral size was measured at multiple positions on preoperative MRI images. A blinded reviewer used three-dimensional CT analysis to record lateral center edge angle (LCEA), acetabular version, Tonnis angle, acetabular coverage, alpha angle, femoral torsion, and neck-shaft angle (FNSA). T2 MRI mapping values of the labrum were obtained via sequencing analyses on each patient’s optimal sagittal cut. Univariate mixed linear models were used to identify associations between each radiographic measurement and decision to repair or reconstruct the labrum. Fifty-two operations were included. Labral size had no predictive effect on undergoing labral reconstruction versus repair. Likelihood for undergoing labral reconstruction was associated with LCEA (P = .003) and Tonnis angle (P = .034). There was an association (P &amp;lt; .05) between labral T2 mapping values and all radiographic parameters except for FNSA and combined version. Labral size was not associated with whether patients underwent labral reconstruction or repair. The data showed an association between labrum T2 mapping values and nearly all radiographic parameters.

MRI without contrast enhancement is a highly accurate method of imaging insulinoma

BACKGROUND: According to the literature, there are contradictory results regarding the diagnostic accuracy of magnetic resonance imaging (MRI) without contrast enhancement (CE) in insulinoma, and studies that analyze the factors influencing the obtaining of false negative results have not been described.AIM: Evaluation of the operational characteristics of MRI imaging of insulinoma without CE.MATERIALS AND METHODS: Imaging studies of the abdominal cavity and retroperitoneal space were performed in patients with suspected nondiabetic hypoglycemia (NDH): MRI without CE (index test) and CT with CE (first row reference test). Ultrasound examination was used as a reference test of the second row, and arteriostimulated venous blood sampling was used in the third row. The diagnosis of NDH and insulinoma was established based on the detection of hyperinsulinemic hypoglycemia during laboratory examination and confirmation of the presence of a pancreatic tumor.RESULTS: 125 patients with insulinoma and 55 patients with other causes of NDH were included. The sensitivity of MRI without CE in the diagnosis of insulinoma was 94%, 95% CI [89%; 98%], specificity — 98%, 95% CI [90%; 100%], prognostic value of a positive result — 99%, 95% CI [95%; 100%], prognostic value of a negative result is 87%, 95% CI [76%; 94%]. The Kappa index of agreement with the final diagnosis was 0.886 (95% CI [0.814; 0.958]), which corresponds to excellent agreement, with CT — 0.750 (95% CI [0.651; 0.850]), which corresponds to significant agreement.In a comparative analysis of the tomographic and pathomorphological characteristics of insulinoma, in cases with divergent results when using MRI without CE and CT with CE, significant differences in structure, shape, contour features, localization and degree of malignancy of the tumor, as well as the structure of its surrounding parenchyma were not revealed.In the presented sample, the insulinoma has a median size of 14–15 mm with the median optical density in the native phase of CT 42–44 units.H, hyperintensivity on T2-weighted MRI images in 89%, 95% CI [78%; 95%] cases and low Grade 1 malignancy in 70%, 95% CI [56%; 81%] cases.CONCLUSION: CT with CE and MRI without CE are characterized by significant agreement in the diagnosis of insulinoma. In this regard, as well as due to the lack of radiation exposure and the strict need for contrast agents, MRI without CE can be recommended as a first-line method on a par with abdominal ultrasound and CT with CE to detect insulinoma.

The need for a second transurethral resection in high-risk non-muscle-invasive bladder cancer based on the Vesicle Imaging-Reporting and Data System.

The efficacy of Vesical Imaging-Reporting and Data System (VI-RADS) for the second transurethral resection (TUR) has not been adequately validated. This study aimed to evaluate the utility of the VI-RADS for high-risk patients with non-muscle-invasive bladder cancer (NMIBC) who are candidates for a second TUR. We retrospectively analyzed 116 patients who received magnetic resonance imaging (MRI) prior to an initial TUR and underwent a second TUR for a diagnosis of high-risk NMIBC at the initial TUR. MRI images were retrospectively classified according to VI-RADS. Second TUR outcomes and recurrence-free and progression-free survival rates were compared with VI-RADS scores. Ninety-nine (91%) patients were diagnosed with T1 bladder cancer at the initial TUR. At the second TUR, residual cancer was found in 53 (49%) cases, including five (4.6%) cases of muscle invasion. With a median follow-up of 41 months, the 2-year bladder recurrence-free survival rate was 71% and the 2-year progression-free rate was 85%. By two radiologists' consensus, 30 (28%)/49 (45%)/16 (15%)/10 (9.2%)/4 (3.7%) cases were classified as VI-RADS 1/2/3/4/5, respectively. Of five pT2 upstage cases, three were VI-RADS 1, one was VI-RADS 2, and one was VI-RADS 3. There was no significant association between VI-RADS and cancer residual rate and pT2 upstage rate in second TUR outcomes, and recurrence-free and progression-free survival rates. In high-risk NMIBCs, a certain number of residual cancers and pT2 upstage cases exist after the initial TUR, and a second TUR should be performed regardless of VI-RADS scores.

An investigation into the applicability of rapid artificial intelligence‐assisted compressed sensing in brain magnetic resonance imaging performed at 5 Tesla field strength

AbstractBackgroundBrain magnetic resonance imaging (MRI) at 5 T offers unprecedented spatial resolution but is often limited by long scan times. Acceleration techniques, such as compressed sensing (CS) and artificial intelligence‐assisted compressed sensing (ACS), have the potential to speed up the acquisition process while maintaining image quality. This study aims to evaluate and compare the performance of CS and ACS (with various acceleration factors) in brain MRI imaging at 5 T.MethodsIn this study, we enrolled 12 healthy volunteers and compared ACS‐accelerated 5 T brain MRI with conventional methods of CS. The ACS acceleration factors for the brain protocol, consisting of 3D T1‐weighted sequences and 2D T2‐weighted sequences, were optimized in a pilot study on healthy volunteers (acceleration factor, 2.06–3.41× in T2‐weighted imaging and 3.52–8.49× in T1‐weighted imaging). We evaluated the images acquired from patients using various acceleration methods on the basis of acquisition times, the signal‐to‐noise ratio (SNR), the contrast‐to‐noise ratio, subjective image quality, and diagnostic agreement.ResultsOur findings revealed that ACS acceleration significantly reduced the acquisition times for T1‐ and T2‐weighted sequences by up to 43% and 53%, respectively, compared with traditional CS at 5 T. Importantly, this acceleration was achieved while maintaining excellent image quality, demonstrated by higher or comparable SNR and contrast‐to‐noise ratio values.ConclusionsThe optimal ACS acceleration factors for 5 T brain MRI were determined to be 2.73× for 2D T2‐weighted sequences and 6.5× for 3D T1‐weighted sequences. ACS not only facilitates rapid imaging but also ensures comparable image quality and diagnostic performance, highlighting its potential to revolutionize high‐field MRI scanning.

Hypoxic brain damage in Methadone misuse: insights from MRI imaging and comparative study.

This study aimed to investigate the potential presence of brain disorders, particularly hypoxia, via magnetic resonance imaging (MRI) in patients misusing methadone, with a comparison to regular opium users and a control group. Conducted as a cross-sectional comparative study at Kamali Hospital in Karaj, Iran, the research included male participants comprising methadone users, opium users, and controls. Inclusion criteria were stringent, focusing on substance use duration and absence of brain structural disorders. MRI scans were performed using a 1.5T MRI scanner. Qualitative MRI assessments and chi-square tests analyzed associations between substance use and hypoxia, while logistic regression examined potential confounding variables. Significant hypoxia was observed in the methadone group (16.7%, 5/24; p = 0.00057), with no cases in the opium or control groups. Logistic regression analysis showed no significant predictors of hypoxia regarding dose and duration of use. MRI findings in methadone users with hypoxia included varied ADC intensities, high signal intensities on T2-weighted and diffusion-weighted imaging (DWI) sequences, and angiogenesis patterns on TOF sequences. The co-use of methadone and alcohol was noted in three of the five hypoxia cases. Methadone misuse, particularly with alcohol, poses a significant risk of hypoxia, detectable via MRI. This study underscores the need for routine MRI monitoring, stricter regulation of non-prescribed methadone, and enhanced public health education to mitigate misuse risks. Future research should expand sample sizes and incorporate advanced imaging techniques to further elucidate methadone's neurological impact.

Magnetic resonance imaging findings of cerebral venous malformations.

Our study aimed to evaluate the incidence and anatomical locations of cerebral venous malformations (CVMs), their alterations over time, and magnetic resonance imaging (MRI) findings of their relationships with each other in patients who underwent magnetic susceptibility-weighted imaging (SWI) and contrast-enhanced MRI. The drainage pattern of developmental venous anomalies (DVAs), the number of collecting veins, DVA thrombosis, signal-intensity abnormalities, and hemosiderin deposits (nodular, diffuse, or both) related to the DVA, were examined. The alterations over time in cavernomas (type and size) were assessed. The study evaluated 7,826 patients who had 8,957 magnetic SWIand contrast-enhanced brain MRIs. A total of 643 CVMs were detected in 520 patients. The frequencies of DVAs, cavernomas, and capillary telangiectasias were 5.26%, 1.28%, and 0.62%, respectively. Isolated DVAs were detected in 430 (71.5%), cavernomas in 77 (12.8%), and capillary telangiectasias in 52 (8.7%) cases of CVMs. The coexistence of DVA with cavernoma and DVA with capillary telangiectasia was found in 38 (6.3%) and 4 (0.7%) cases, respectively. DVA drainage was largely provided by a single collecting vein. Signal-intensity abnormalities associated with the DVA were observed in 7, and two patients had DVA thrombosis. Hemosiderin deposits were nodular in 14 cases, diffuse in three, and both nodular and diffuse in two. In a follow-up of 11 (7.8%) cavernomas, alterations in the type and size were detected. It is not uncommon for cavernomas to become symptomatic owing to alterations in their nature, size, and type over time, while DVAs rarely become symptomatic. Capillary telangiectasias are vascular malformations that tend to remain stable in terms of clinical and imaging features, and those with supratentorial localization are detected more frequently in the frontal lobe. On magnetic susceptibility sequences, hemosiderin deposits can be seen in the GVA region, not rarely but more frequently in nodular form, as well as in diffuse or nodular or diffuse and nodular forms.

Wavelets-Based Texture Analysis of Post Neoadjuvant Chemoradiotherapy Magnetic Resonance Imaging as a Tool for Recognition of Pathological Complete Response in Rectal Cancer, a Retrospective Study.

Background: Patients with locally advanced rectal cancer (LARC) treated by neoadjuvant chemoradiotherapy (nCRT) may experience pathological complete response (pCR). Tools that can identify pCR are required to define candidates suitable for the watch and wait (WW) strategy. Automated image analysis is used for predicting clinical aspects of diseases. Texture analysis of magnetic resonance imaging (MRI) wavelets algorithms provides a novel way to identify pCR. We aimed to evaluate wavelets-based image analysis of MRI for predicting pCR. Methods: MRI images of rectal cancer from 22 patients who underwent nCRT were captured at best representative views of the tumor. The MRI images were digitized and their texture was analyzed using different mother wavelets. Each mother wavelet was used to scan the image repeatedly at different frequencies. Based on these analyses, coefficients of similarity were calculated providing a variety of textural variables that were subsequently correlated with histopathology in each case. This allowed for proper identification of the best mother wavelets able to predict pCR. The predictive formula of complete response was computed using the independent statistical variables that were singled out by the multivariate regression model. Results: The statistical model used four wavelet variables to predict pCR with an accuracy of 100%, sensitivity of 100%, specificity of 100%, and PPV and NPV of 100%. Conclusions: Wavelet-transformed texture analysis of radiomic MRI can predict pCR in patients with LARC. It may provide a potential accurate surrogate method for the prediction of clinical outcomes of nCRT, resulting in an effective selection of patients amenable to WW.

A Pilot Study of PSMA PET/CT and MRI Fusion for Prostate Cancer: Software to Replace PET/MRI Hardware.

Introduction: Prostate-specific membrane antigen positron emission tomography/computed tomography (PSMA PET/CT), in combination with magnetic resonance imaging (MRI), may enhance the diagnosis and staging of prostate cancer. Image fusion of separately acquired PET/CT and MRI images serve to facilitate clinical integration and treatment planning. This study aimed to investigate different PSMA PET/CT and MRI image fusion workflows for prostate cancer visualisation. Methods: Eighteen patients with prostate cancer who underwent PSMA PET/CT and MRI prior to radical prostatectomy were retrospectively selected. Alignment of the prostate was performed between PET/CT and MRI via three techniques: semi-automatic rigid, automatic rigid, and automatic non-rigid. Image fusion accuracy was evaluated through boundary and volume agreement, quantified by the Dice Similarity Coefficient (DSC), 95% Hausdorff Distance (HD), and Mean Surface Distance (MSD), with comparison against reconstructed histopathology slices. Results: Image fusion using all techniques resulted in clear lesion visualisation from PSMA PET/CT overlay and anatomical detail afforded by the MRI base and was consistent with histopathology tumour location. Image fusion accuracy was within the recommended range based on a DSC of 0.8-0.9. The automatic non-rigid registration method had the highest volume agreement (DSC: 0.96 ± <0.01) and boundary agreement (HD: 1.17 ± 0.35 mm) when compared to automatic rigid (DSC 0.88 ± 0.02, HD 3.18 ± 0.29 mm) and semi-automatic rigid (DSC 0.80 ± 0.06, HD 5.25 ± 1.68 mm). Conclusions: Image fusion of clinically obtained PET/CT and MRI is feasible and clinically acceptable for use in prostate cancer diagnosis and surgical management. While the best accuracy was observed with the automatic non-rigid technique, which requires further validation, image fusion with clinically accessible methods (semi-automatic rigid) may currently aid patient education, pre-operative planning, and intra-operative guidance.

Multimodal MRI Deep Learning for Predicting Central Lymph Node Metastasis in Papillary Thyroid Cancer.

Central lymph node metastasis (CLNM) in papillary thyroid cancer (PTC) significantly influences surgical decision-making strategies. This study aims to develop a predictive model for CLNM in PTC patients using magnetic resonance imaging (MRI) and clinicopathological data. By incorporating deep learning (DL) algorithms, the model seeks to address the challenges in diagnosing CLNM and reduce overtreatment. The results were compared with traditional machine learning (ML) models. In this retrospective study, preoperative MRI data from 105 PTC patients were divided into training and testing sets. A radiologist manually outlined the region of interest (ROI) on MRI images. Three classic ML algorithms (support vector machine [SVM], logistic regression [LR], and random forest [RF]) were employed across different data modalities. Additionally, an AMMCNet utilizing convolutional neural networks (CNNs) was proposed to develop DL models for CLNM. Predictive performance was evaluated using receiver operator characteristic (ROC) curve analysis, and clinical utility was assessed through decision curve analysis (DCA). Lesion diameter was identified as an independent risk factor for CLNM. Among ML models, the RF-(T1WI + T2WI, T1WI + T2WI + Clinical) models achieved the highest area under the curve (AUC) at 0.863. The DL fusion model surpassed all ML fusion models with an AUC of 0.891. A fusion model based on the AMMCNet architecture using MRI images and clinicopathological data was developed, effectively predicting CLNM in PTC patients.

Automatic deep learning method for analysis and prediction of neonatal hyperbilirubinemia in magnetic resonance imaging.

This study conducted a comparative analysis among newborns with varying levels of hyperbilirubinemia, explored the relationships between magnetic resonance imaging (MRI) image features and serum bilirubin levels in hyperbilirubinemia, and proposed an automatic classification system based on deep learning (DL) for prediction of neonatal hyperbilirubinemia (NHB). This retrospective study enrolled 606 consecutive neonates who had their serum bilirubin detected at the Xi'an Fourth Hospital, including 273 cases of patients and 333 cases of normal controls. After data preprocessing, MRI images were fed into the Inception-v3 network, graph convolutional network (GCN), and 3-dimensional (3D) patch-based GCN that introduced the graph attention mechanism (our GCN) for NHB analysis and classification, respectively. Multi-threshold grouping was conducted based on various serum bilirubin levels. Performance evaluation involved the area under the curve (AUC), accuracy (ACC), sensitivity (SEN), and specificity (SPE). As the bilirubin levels gradually increased, the overall performance metrics of DL system for detecting the T1-weighted imaging signal in the pallidum region showed a significant upward trend. Our GCN for the prediction and classification of MRI image features of NHB achieved satisfactory results. When the bilirubin value exceeded 400 µmol/L, it achieved an AUC of 0.86 and ACC of 0.81, which is significantly higher than other advanced models (ACC: 72-78.3%) with the same proposed input form. The DL system has the potential to automatically analyze and predict NHB on MRI.

St-RegSeg: an unsupervised registration-based framework for multimodal magnetic resonance imaging stroke lesion segmentation.

Stroke is one of the leading causes of disability and death worldwide. Ischemic stroke accounts for 75-90% of all stroke incidents. Assessing the size and location of the stroke lesion is crucial for treatment decisions, especially those related to urgent vascular reconstruction surgery. Magnetic resonance imaging (MRI) offers excellent soft tissue contrast and multimodal imaging characteristics, which can reflect changes in the physiological functions of brain soft tissues in patients with ischemic stroke. However, using deep learning (DL) techniques for MRI segmentation of stroke lesions still faces many challenges. On the one hand, single-modal segmentation models cannot effectively integrate multimodal information; on the other hand, there is a semantic content drift between multimodal stroke MRI images, leading to lower accuracy in subsequent multimodal image segmentation. To address these issues, we aimed to propose the stroke unsupervised registration and segmentation (St-RegSeg) framework. The St-RegSeg framework integrates an unsupervised registration model, ConvNXMorph, and a segmentation model, nnUNet-v2, enabling both registration and segmentation of multimodal MRI images. The St-RegSeg framework was evaluated on the ISLES'22 dataset from three centers. The St-RegSeg framework demonstrated significant improvements in performance metrics and computational efficiency. Compared to advanced normalization tools (ANTs) [symmetric normalization (SyN)] + nnUNet-v2, the St-RegSeg framework improved the Dice similarity coefficient (DSC) by 25.31% in the registration phase, reduced mean squared error (MSE) by 17.36%, increased normalized cross-correlation (NCC) by 16.06%, and enhanced mutual information (MI) by 17.09%. Additionally, in the segmentation phase, it increased the DSC by 0.84%, and the overall inference speed was increased by 40.91 times. Compared to the suboptimal TransMorph + nnUNet-v2, the St-RegSeg framework improved the DSC by 3.68% in the registration phase, reduced MSE by 8.91%, increased NCC by 8.49%, enhanced MI by 6.18%, and in the segmentation phase, it raised the DSC by 0.5%, with the overall inference speed increased by 2.13 times. The St-RegSeg framework provides a highly effective solution for the registration and segmentation of multimodal MRI images in ischemic stroke cases. Its performance metrics and computational efficiency significantly outperform existing methods, making it a promising tool for clinical applications. The code is open-sourced and available at: https://github.com/Cooper-Gu/St-RegSeg.

Redefining Ventricular Access Landmarks: A Novel Approach Using MRI Imaging and 3D Reconstructions for Keen’s Point Trajectories

IntroductionCerebrospinal Fluid (CSF) diversion via shunt placement is a common neurosurgical intervention. Keen’s Point is a widely used landmark for accessing the trigone of the lateral ventricle; however, it lacks clinical validation. This study re-evaluates Keen’s Point scalp parameters using magnetic resonance imaging (MRI) to identify an ideal shunt entry point based on 3D reconstructions. MethodsMRI data from 36 Parkinson’s disease patients scheduled for Deep Brain Stimulation (DBS) were analyzed. The optimal ventricular puncture site (A Point) was identified on axial MRI images and mapped to the scalp. Distances from A Point to a perpendicular line from the external auditory meatus (EAM; B Point) and from B Point to the pinna and EAM were measured. The Posterior Callosal Angle (PCA) was calculated to assess its correlation with the distance between A and B points. Statistical analysis used regression and t-tests (p < 0.05). ResultsAfter excluding patients with PCA outside 70–100 degrees, 24 patients remained. The optimal distance from B Point to A Point significantly increased from 25–30 mm (as described by Keen) to 56.89 ± 1.36 mm (p < 0.01). Average PCA angle (85 degrees) correlates with a distance of ∼57 mm, and a Significant negative correlation was found between PCA and the distance from A Point to B Point (r = -0.42, p < 0.03). ConclusionThis modified Keen’s Point is a reliable landmark for ventriculostomy. Incorporating PCA into preoperative planning enhances catheter placement accuracy, providing a prominent approach to shunting at Keen’s Point.

Non‐Destructive Prediction Approach for Pomelo Granulation Rate Using MRI Technology

ABSTRACTPomelo is prone to juice sac granulation during ripening and storage, resulting in decreased moisture content, hard flesh, and bland flavor, which affects market value and consumer experience. Due to the complexity of the causes of juice sac granulation, coupled with the thick peel of pomelo, the detection means of pomelo is lacking and difficult to detect. In order to solve the shortcomings of the traditional granulation rate calculation, which has an error and cannot realize non‐destructive detection. This study proposes a non‐destructive detection of pomelo juice sac granulation and the calculation of granulation rate based on the nuclear magnetic resonance imaging (MRI) technology combined with the computer vision technology. Based on the strong analytical power of the MRI technology for moisture and the change of moisture in the granulated juice sac of pomelo, in this study, the cross section of pomelo was scanned by MRI scanner to obtain the corresponding MRI images. The MRI images of the cross section of pomelo were segmented using the threshold segmentation method to realize the calculation of the percentage of granulated pulp. The experimental results show that the segmentation threshold T1 = 102 can effectively segment the pulp from the peel, and the segmentation threshold T2 = 175 can effectively segment the granulated pulp from the normal pulp. The granulation rate of pomelo pulp can be calculated by calculating the ratio of the granulated pulp area to the total pulp area. Therefore, based on the MRI technology can be used for the detection of the granulation rate of pomelo juice sac, which provides a certain reference value for the subsequent non‐destructive detection of the internal quality of large melons and fruits.