Current Medical Imaging - Online First

The aim of this study was to develop and validate predictive models for perineural invasion (PNI) in gastric cancer (GC) using clinical factors and radiomics features derived from contrast-enhanced computed tomography (CE-CT) scans and to compare the performance of these models.

This study included 205 GC patients, who were randomly divided into a training set (n=143) and a validation set (n=62) in a 7:3 ratio. Optimal radiomics features were selected using the least absolute shrinkage and selection operator (LASSO) algorithm. A radiomics model was constructed utilizing the optimal among five machine learning filters, and a radiomics score (rad-score) was computed for each participant. A clinical model was built based on clinical factors identified through multivariate logistic regression. Independent clinical factors were combined with the rad-score to create a combined radiomics nomogram. The discrimination ability of the models was evaluated by receiver operating characteristic (ROC) curves and the DeLong test.

Independent predictive factors of the clinical model included tumor T stage, N stage, and tumor differentiation, with AUC values of 0.777 and 0.809 in the training and validation sets. The radiomics model was constructed using the support vector machine (SVM) classifier with the best AUC (0.875 in the training set and 0.826 in the validation set). The combined radiomics nomogram, which combines independent clinical predictors and the rad-score, demonstrated better predictive performance (AUC=0.889 in the training set; AUC=0.885 in the validation set).

The nomogram integrating independent clinical predictors and CE-CT radiomics was constructed to predict PNI in GC. This model demonstrated favorable performance and could potentially assist in prognosis evaluation and clinical decision-making for GC patients.

This study aimed to assess the hemodynamic changes in the vena cava and predict the likelihood of Cardiac Remodeling (CR) and Myocardial Fibrosis (MF) in athletes utilizing four-dimensional (4D) parameters.

A total of 108 athletes and 29 healthy sedentary controls were prospectively recruited and underwent Cardiac Magnetic Resonance (CMR) scanning. The 4D flow parameters, including both general and advanced parameters of four planes for the Superior Vena Cava (SVC) and Inferior Vena Cava (IVC) (sheets 1-4), were measured and compared between the different groups. Four machine learning models were employed to predict the occurrence of CR and/or MF.

Most general 4D flow parameters related to VC were increased in athletes and positive athletes compared to controls (p < 0.05). Gradient Boosting Machine (GBM) was the most effective model in sheet 2 of SVC, with the area under the curve values of 0.891, accuracy of 85.2%, sensitivity of 84.6%, and specificity of 85.4%. The top five predictors in descending order were as follows: net positive volume, forward volume, waist circumference, body weight, and body surface area.

Physical activity can induce a high flow state in the vena cava. CR and/or MF may elevate the peak velocity and maximum pressure gradient of the IVC. This study successfully constructed a GBM model with high efficacy for predicting CR and/or MF. This model may provide guidance on the frequency of follow-up and the development of appropriate exercise plans for athletes.

Multiple sclerosis (MS) is one of the most common disabling central nervous system diseases affecting young adults. Magnetic resonance imaging (MRI) is an essential tool for diagnosing and following up multiple sclerosis. Over the years, many MRI techniques have been developed to improve the sensitivity of MS disease detection. In recent years synthetic MRI (sMRI) and quantitative MRI (qMRI) have gained traction in neuroimaging applications, providing more detailed information than traditional acquisition methods. These techniques enable the detection of microstructural changes in the brain with high sensitivity and robustness to inter-scanner and inter-observer variability. This study aims to evaluate the feasibility of using these techniques to avoid administering intravenous gadolinium-based contrast agents (GBCAs) for assessing MS disease activity and monitoring.

Forty-two known MS patients, aged 20 to 45, were scanned as part of their routine follow-up. MAGnetic resonance image Compilation (MAGiC) sequence, an implementation of synthetic MRI, was added to our institute's routine MS protocol to automatically generate quantitative maps of T1, T2, and PD. T1, T2, PD, and apparent diffusion coefficient (ADC) data were collected from regions of interest (ROIs) representing normal-appearing white matter (NAWM), enhancing, and non-enhancing MS lesions. The extracted information was compared, and statistically analyzed, and the sensitivity and specificity were calculated.

The mean R1 (the reciprocal of T1) value of the non-enhancing MS lesions was 0.694 s^-1 (T1=1440 ms), for enhancing lesions 1.015 s^-1 (T1=985ms), and for NAWM 1.514 s^-1 (T1=660ms). For R2 (the reciprocal of T2) values, the mean value was 6.816 s^-1 (T2=146ms) for non-enhancing lesions, 8.944 s⁻¹ (T2=112 ms) for enhancing lesions, and 1.916 s⁻¹ (T2=522 ms) for NAWM. PD values averaged 93.069% for non-enhancing lesions, 82.260% for enhancing lesions, and 67.191% for NAWM. For ADC, the mean value for non-enhancing lesions was 1216.60×10⁻⁶ mm²/s, for enhancing lesions 1016.66×10⁻⁶ mm²/s, and for NAWM 770.51×10⁻⁶ mm²/s.

Our results indicate that enhancing and non-enhancing MS lesions significantly decrease R1 and R2 values. Non-enhancing lesions have significantly lower R1 and R2 values compared to enhancing lesions.

Conversely, PD values are significantly higher in non-enhancing lesions than in enhancing lesions. For ADC, while NAWM has lower values, there was minimal difference between the mean ADC values of enhancing and non-enhancing lesions.

Early and timely detection of pulmonary nodules and initiation treatment can substantially improve the survival rate of lung carcinoma. However, current detection methods based on convolutional neural networks (CNNs) cannot easily detect pulmonary nodules owing to low detection accuracy and the difficulty in detecting small-sized pulmonary nodules; meanwhile, more accurate CNN-based models are slow and require high hardware specifications.

The aim of this study is to develop a detection model that achieves both high accuracy and real-time performance, ensuring effective and timely results.

In this study, based on YOLOv5s, a concentrated-comprehensive convolution (C3_ODC) module with multidimensional attention is designed in the convolutional layer of the original backbone network for enhancing the feature-extraction capabilities of the model. Moreover, lightweight convolution is combined with weighted bidirectional feature pyramid networks (BiFPNs) to form a GS-BiFPN structure that enhances the fusion of multiscale features while reducing the number of model parameters. Finally, Focal Loss is combined with the normalized Wasserstein distance (NWD) to optimize the loss function. Focal loss focuses on carcinoma-positive samples to mitigate class imbalance, whereas the NWD enhances the detection performance of small lung nodules.

In comparison experiments against the YOLOv5s, the proposed model improved the average precision by 8.7% and reduced the number of parameters and floating-point operations by 5.4% and 8.2%, respectively, while achieving 116.7 frames per second.

The proposed model balances high detection accuracy against real-time requirements.

Perforation is one of the rarest effects of Meckel's diverticulum and may clinically resemble acute appendicitis.

A 34-year-old woman with pain in the right iliac fossa, nausea, and vomiting for three days was brought to the emergency department. An abdominal examination indicated rebound tenderness in the area of the right iliac fossa. Abdominal ultrasound showed a heterogeneous lesion in the left iliac fossa measuring 5 cm × 3.5 cm × 4 cm with no internal vascularity. Abdominal Computed Tomography (CT) demonstrated a hypodense lesion located left of the midline of the abdomen, which was inseparable from the small bowel at the antimesenteric border. Laparoscopic exploration was performed, and an intraoperative diagnosis of perforated Meckel’s diverticulum with phlegmon formation was made. The patient had an uneventful recovery.

Radiologists should be aware of the possibility of complicated Merkel's diverticulum when encountering cases of acute abdominal pain. If there is a lower abdominal inflammatory process and a normal appendix is identified, there should be a high degree of suspicion when examining the CT scan.

Fibrosing mediastinitis (FM) is a rare and benign disease affecting the mediastinum and often causes pulmonary hypertension (PH). Timely diagnosis of PH caused by FM is clinically important to mitigate complications such as right heart failure in affected individuals. This retrospective study aimed to analyze the CT imaging characteristics of TB-related FM in patients with tuberculosis (TB). Additionally, the study investigates the underlying reasons contributing to the manifestation of symptoms.

From April 2007 to October 2020, high-resolution CT (HRCT) and dual-phase CT pulmonary angiography images of 64 patients with suspected FM diagnosed with PH at a tertiary hospital were examined. The imaging characteristics of these CT scans were analyzed, with a specific focus on the TB-FM involvement of the pulmonary veins, pulmonary arteries, and bronchi (down to the segment level).

HRCT imaging revealed that fibrous tissue inside the mediastinum exhibited minimal or negligible reinforcement in TB-FM and diffuse fibrous infiltration in the mediastinum and hilar areas. Notably, segmental bronchial and pulmonary artery stenosis are more pronounced and frequently co-occurring than lobe-level stenosis. Pulmonary venous stenosis developed outside the pericardium, whereas pulmonary artery stenosis occurred outside the mediastinal pleura. Furthermore, no isolated FM involvement in pulmonary veins was noticed in this cohort.

HRCT imaging of TB-related FM presents unique features in certain regions of the bronchi, pulmonary veins, and pulmonary arteries. Thus, it is imperative to accurately identify fibrous tissue involvement in mediastinal lesions for proper diagnosis and management of TB-FM.

Pneumonia is an acute respiratory infection that has emerged as the predominant catalyst for escalating mortality rates worldwide. In the pursuit of the prevention and prediction of pneumonia, this work employs the development of an advanced deep-learning model by using a federated learning framework. The deep learning models rely on the utilization of a centralized system for disease prediction on the medical imaging data and pose risks of data breaches and exploitation; however, federated learning is a decentralized architecture which significantly reduces data privacy concerns.

The federated learning works in a distributed architecture by sending a global model to clients rather than sending the data to the model. The proposed federated deep learning-based FedPneu computer-aided diagnosis model has been implemented in 2, 3, 4, and 5 clients architecture for early pneumonia detection using X-ray images. The key parameters configuration include batch size, learning rate, optimizer, decay, momentum, epochs, rounds, and random-split as 32, 0.0001, SGD, 0.000001, 0.9, 10, 100, and 42, respectively.

The results of the proposed federated deep learning-based FedPneu model have been provided in terms of round-wise accuracy, loss, and computational time. The highest accuracy of 85.632% has been achieved with 2-clients federated deep learning architecture, whereas, 3, 4, and 5 clients architecture achieved 85.536%, 76.112%, and 74.123% accuracies, respectively.

In the proposed privacy-protected federated deep learning-based FedPneu model, the two-client architecture has been resulted as the most optimal framework for pneumonia detection among 3-clients, 4-clients, and 5-clients architecture. The model works in a collaborative and privacy-protected framework with a multi-silo dataset which could be highly beneficial for healthcare departments to maintain patient’s data privacy with improved prediction outcomes.

This prospective study aimed to establish the typical viscosity range of the thyroid gland in healthy individuals using a new method called the Sound Touch Viscosity (STVi) technique with a linear array transducer.

Seventy-eight healthy volunteers were enrolled between March, 2023 and April, 2023. Thyroid viscosity was measured using the Resona R9 ultrasound system equipped with a linear array transducer (L15-3WU). Each patient had three valid viscosity measurements taken for each thyroid lobe, and the average values were analyzed. Thyroid gland stiffness was measured and analyzed simultaneously.

The study included 51 women and 27 men with an average age of 48 years. The mean viscosity measurement for a normal thyroid gland was 1.10 ± 0.41 Pa.s (ranging from 0.38 to 2.25 Pa.s). There were no significant differences in viscosity between the left and right lobes of the thyroid gland. We found no significant variations in viscosity based on gender, age, or body mass index (BMI). There was a notable positive correlation between thyroid viscosity and stiffness measurements (r = 0.717, p < 0.001).

Our findings suggest that STVi is a highly reliable method for assessing the thyroid. This technique holds promise as a new, non-invasive approach to evaluating thyroid parenchyma viscosity.

Mucinous Cystadenocarcinoma (MCA) of the breast remains a relatively rare condition, and to date, there is no systematic summary of its imaging manifestations. Therefore, this report presents a detailed account of the diagnosis and treatment of mucinous cystadenocarcinoma in a 40-year-old woman, with a particular focus on imaging findings. Additionally, we conducted a comprehensive literature review on this disease and summarized its key imaging features. This manuscript provides valuable insights and methodologies for the accurate diagnosis of mucinous cystadenocarcinoma.

We report a 40-year-old premenopausal woman who discovered multiple cysts in her left breast five years ago. Over the past two years, the size of these tumors has increased. Ultrasound examination indicated that the cysts had grown to 27 x 17mm. Following a puncture, the cysts were confirmed to be benign and were not monitored regularly. A year later, the patient's mass in the left breast increased, and an ultrasound exam indicated a suspicious mixed echo area in the upper outer quadrant, suggestive of a malignant lesion. Mammography showed amorphous suspicious calcifications in the lesion area, distributed in segments. Contrast-enhanced magnetic resonance imaging displayed non-mass-type enhancement of the lesion, with a dynamic enhanced imaging time-signal intensity curve (TIC) showing a rapidly rising plateau pattern. Postoperative pathology confirmed invasive carcinoma of the left breast along with mucinous cystadenocarcinoma. Four months after surgery, the patient developed multiple abnormal lymph nodes in the left axilla, which were confirmed to be metastasis upon pathology examination. Following radiotherapy, the patient's condition remained stable during the follow-up period.

Most MCA lesions typically exhibit clear borders and irregular edges, with some displaying expansive growth and compression of surrounding tissues. Mammography can reveal calcified components in lesions. Ultrasound often reveals an isoechoic or hypoechoic mass with well-defined borders but irregular edges. Magnetic resonance imaging (MRI) can show clear boundaries and uneven enhancement of the lesions, and the time-intensity curve (TIC) of the mass area often shows an inflow enhancement pattern.