Neoadjuvant therapy is the core treatment strategy for locally advanced breast cancer, and its efficacy evaluation directly determines surgical planning, prognostic stratification, and subsequent therapeutic regimens. Leveraging the strengths of multi-parametric and multi-sequential imaging, advanced techniques have moved beyond traditional morphological assessment to the non-invasive quantification of the tumor microenvironment. These innovations have become pivotal tools for assessing response to neoadjuvant therapy in breast cancer. This review synthesizes the current applications of advanced MRI techniques, analyzes their technical advantages, and outlines future research directions, with the aim of promoting their deeper integration into breast cancer care and contributing to the refinement of precision diagnosis and treatment.
Objective To investigate the value of ultrafast dynamic contrast-enhanced MRI (DCE-MRI) performed before treatment and after the early treatment stage in predicting the efficacy of neoadjuvant chemotherapy (NAC) for breast cancer. Methods A total of 101 female patients with breast cancer undergoing NAC were prospectively enrolled (mean age, 49±8 years). Based on postoperative pathological results, patients were divided into the pathological complete response (pCR) group (26 cases) and the non-pCR group (75 cases). All patients underwent ultrafast DCE-MRI and diffusion-weighted imaging (DWI) both before treatment and after two treatment cycles. Clinicopathological characteristics were recorded, and ultrafast DCE-MRI parameters and apparent diffusion coefficient (ADC) values were measured at both time points. Differences in clinicopathological variables and MRI parameters between groups were analyzed using the Chi-square test, independent-sample t test, or Mann-Whitney U test. Multivariate logistic regression analysis was performed to identify independent predictors of pCR and to construct prediction models. The predictive performance of each model was evaluated using receiver operating characteristic (ROC) and the area under the curve (AUC), with DeLong test used for comparisons between models. Results Significant differences were observed between the two groups in ER, PR, HER2, Ki67, lymph node metastasis, and clinical stage (all P<0.05). Compared with baseline, Ktrans, wash in slope(WIS), peak enhancement intensity (PEI), and initial area under the curve in 60 s(iAUC) values after two treatment cycles were significantly decreased, while time-to-peak(TTP) was significantly increased (all P<0.05). Multivariate Logistic regression analysis identified PR status, post-treatment PEI, and ΔWIS as independent predictors of pCR. Four models were established based on different predictors: a clinicopathological model, an ultrafast DCE-MRI model, fusion model 1 (combining post-treatment ADC value and PR), and fusion model 2 (integrating PR, post-treatment PEI, and ΔWIS). The diagnostic performance of fusion model 2 (AUC=0.887) was superior to that of the clinicopathological model (AUC=0.702, Z=5.398, P<0.001) and fusion model 1 (AUC=0.764, Z=2.561, P=0.011). Conclusion Ultrafast DCE-MRI combined with clinicopathological features significantly improves the early prediction of NAC in breast cancer.
Objective To investigate a model based on MRI habitat imaging analysis of axillary lymph node (ALN) heterogeneity combined with clinicopathological features for assessing the response to neoadjuvant therapy (NAT) in breast cancer patients with ALN positivity (ALN+). Methods A total of 369 female patients with pathologically confirmed ALN+ breast cancer were retrospectively enrolled and randomly divided into a training set (n=259) and a validation set (n=110) in a 7∶3 ratio. All patients underwent dynamic contrast-enhanced MRI before treatment. Stable 3D radiomic features were used, and the optimal number of clusters was determined using Gaussian mixture modeling and the Bayesian information criterion to generate habitat imaging and extract subregional features. Four support vector machine (SVM)-based models were constructed: a clinical model, a habitat radiomics model, an ALN heterogeneity score model, and a late-fusion combined model. Predictive performance was evaluated using receiver operating characteristic (ROC) curve analysis, and the area under the curve (AUC) values were compared using the Delong test. Clinical utility was assessed using decision curve analysis. Results The clinical model was constructed based on PR status, HER2 status, and Ki-67 index. The habitat radiomics model was developed using 17 non-zero radiomic features extracted from four subregions. An ALN heterogeneity score model and a combined model were also established. In the validation set, the AUCs of the clinical model, habitat radiomics model, ALN heterogeneity score model, and combined model were 0.79, 0.60, 0.69, 0.85, respectively. The Delong test showed that the AUC of the combined model was significantly higher than those of all individual models (all P<0.05). Decision curve analysis demonstrated that the combined model consistently provided a high net benefit within the threshold probability range of 0.2-0.8, indicating favorable clinical applicability. Conclusion A combined model integrating MRI-based habitat radiomics features, ALN heterogeneity scores, and clinicopathological characteristics may assist in evaluating post-NAT lymph node status and support individualized clinical decision-making.
The hypoxic tumor microenvironment is a key factor affecting the efficacy of neoadjuvant therapy (NAT) in breast cancer, and its heterogeneity complicates treatment outcomes. Functional magnetic resonance imaging (fMRI) techniques enable noninvasive quantification of tumor hypoxia, facilitating hypoxia assessment and therapy optimization. This approach helps to avoid ineffective chemotherapy toxicity, reduce adverse reactions, and guide treatment adjustment. This review summarizes the mechanisms of hypoxia, relevant fMRI principles, and the applications of fMRI-based hypoxia assessment, monitoring, and therapeutic assistance in breast cancer NAT.
Neoadjuvant therapy (NAT) is a common treatment for breast cancer, and MR diffusion imaging has become a critical imaging tool for evaluating the NAT efficacy. MR diffusion imaging encompasses conventional diffusion-weighted imaging (DWI) and various derivative techniques, including diffusion tensor imaging (DTI), intra voxel incoherent motion (IVIM), and diffusion kurtosis imaging (DKI). These techniques can not only capture changes in tissue microstructure but also provide richer and more specific quantitative information regarding tissue perfusion and heterogeneity, enabling early prediction of treatment response to NAT, assessment of residual lesions, prognostic prediction, and evaluation of axillary lymph node status in breast cancer. This review summarizes the research progress and clinical application of MR diffusion imaging in NAT for breast cancer.
The occurrence, progression, and metastasis of breast cancer are closely associated with the tumor microenvironment. Habitat imaging (HI) can characterize the tumor microenvironment by reflecting intratumoral heterogeneity (ITH), and it holds significant value in assisting the evaluation of molecular characteristics, subtype differentiation, predicting tumor treatment response, assessing metastasis risk, and prognostic prediction in breast cancer. This article briefly explains the principles and methods of HI, reviews the research progress of HI in breast cancer, and analyzes current challenges and future directions, aiming to provide references for the precise diagnosis and personalized treatment of breast cancer.
Precise diagnosis and treatment of breast cancer are crucial for improving patient survival rate. Magnetic nanoparticles integrate targeted multimodal imaging and theranostics to obtain detailed imaging information, enabling targeted and synergistic treatment. This approach can enhance therapeutic efficacy while reducing the toxic side effects of systemic therapy, holding great promise for the diagnosis and treatment of breast cancer. This review summarizes recent progress in research on magnetic nanoparticles in multimodal imaging and treatment for breast cancer.
The rapid development of artificial intelligence (AI) technology has enabled increasingly precise segmentation of brain MRI in fetuses, infants, and children. In recent years, AI-based segmentation methods for structural and functional MRI during the early stages of brain development, from fetus to childhood, have advanced continuously. Techniques such as machine learning, convolutional neural networks (CNNs), transformers, and general foundation models have achieved remarkable improvements in segmentation accuracy, efficiency, and robustness, and have shown great potential for clinical applications. This review summarizes recent advances in AI-based segmentation of fetal, infant, and pediatric brain MRI, highlights current limitations and challenges, and discusses prospects for future development.
Fetal and childhood periods are critical stages of early brain development. The advancement of multimodal MRI technologies has provided important tools for investigating brain development and neurodevelopmental disorders. Structural MRI can reveal brain volume and cortical developmental patterns; diffusion MRI can delineate white matter tracts and the development of neural networks; blood oxygenation level dependent (BOLD) functional MRI can elucidate the formation and abnormalities of functional networks; and MRS can reflect the metabolic states of the brain. This review summarizes recent progress in the use of multimodal MRI to investigate fetal and pediatric brain development, as well as its applications in the early diagnosis, disease subtyping, and exploration of comorbidity mechanisms of neurodevelopmental disorders such as attention-deficit/hyperactivity disorder, autism spectrum disorder, and Tourette syndrome.
Ventriculomegaly (VM) is the most common fetal central nervous system abnormality, and fetal brain MRI has become an important modality for evaluating fetal VM. Structural MRI can quantitatively assess brain volume, specific brain regions, and alterations in sulcation and gyrification in fetuses with isolated VM (IVM), whereas functional MRI can reflect microstructural changes, providing information on brain development and white matter fiber tracts in IVM fetuses, thereby as a reference for prenatal assessment. This article reviews research progress on the use of MRI for central nervous system assessment in fetuses with IVM.
Objective To compare the capability of functional quantitative susceptibility mapping (fQSM) and blood oxygen level dependent (BOLD) techniques in detecting brain activation during finger-tapping task functional MRI (fMRI). Methods Twenty-four healthy adult volunteers were prospectively recruited, and fMRI magnitude and phase images of the finger-tapping block task were synchronously acquired by a 3.0 T magnetic resonance imaging scanner. A general linear model (GLM) and independent component analysis (ICA) were used to derive task-evoked activation maps for BOLD and fQSM. Correlation analysis was used to investigate the similarity between time series of BOLD and fQSM fMRI activation components. Paired t-tests or permutation test were used to compare the dynamic range and fractional amplitudes of low-frequency fluctuations of the activation components between BOLD and fQSM. Results GLM showed that BOLD activation was mainly located in the bilateral precentral and postcentral gyri, cerebellar motor areas, while fQSM did not detect any statistically significant activation (voxel-level P<0.001, cluster-level FWE corrected P<0.05). Even at a lower threshold, activation detected by fQSM was substantially less than that detected by BOLD, and the time series of cerebellar motor areas showed a significant negative correlation between fQSM and BOLD. Head motion and the spatial smoothing levels had no significant effect on the activation intensity of either BOLD or fQSM. ICA with a fixed components (90) showed that BOLD could sensitively detect independent components closely related to the finger tapping task, whereas fQSM failed to identify task-related activation. The correlation between the time series of BOLD and fQSM activation components was weak, and the dynamic range and fractional amplitude of low frequeng fluctuation of BOLD were significantly higher than those of fQSM (P<0.05, Bonferroni correction). Additionally, ICA with automatic component selection identified three fQSM activation components significantly associated with motor function, including the sensorimotor, medial prefrontal, and cerebellar motor networks, with the cerebellar motor network showing a significant negative correlation with the BOLD sensorimotor network. Conclusion The fQSM technique is less sensitive than the conventional BOLD technique in detecting brain activation related to finger movements. ICA with automated component selection may help improve the detectability of fQSM brain activation.
Objective To compare the clinical value of true electrocardiogram (ECG)-gating versus virtual ECG-gating in coronary computed tomography angiography (CCTA). Methods A total of 108 patients with suspected coronary artery disease undergoing CCTA were prospectively enrolled. Because switching between the two gating modes required additional time, participants were alternately assigned by the examination date to the true ECG-gating group (n=54) and the virtual ECG-gating group (n=54). Scans were performed on a United Imaging uCT 960+ wide-detector CT. Examination time and radiation dose were documented. Using the American Heart Association 18-segment model, 5-point subjective image quality scores were obtained, and coronary artery CT attenuation values, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) were measured. For patients who underwent invasive coronary angiography (ICA) within 30 days after CCTA, diagnostic concordance was assessed with ICA as the gold standard, with subgroup analysis by scanning mode. Continuous variables were compared using the t-test or Mann-Whitney U test, and categorical variables with the chi-square test. Receiver operating characteristic (ROC) curve analysis was used to evaluate diagnostic performance. Results In the virtual ECG-gating group, the distal left circumflex artery (LCX) CT attenuation, SNR, and CNR, as well as the subjective image quality score of the posterior descending branch of the right coronary artery, were superior to the true-gating group (all P<0.05). No statistically significant differences were found for other image quality parameters between groups (all P>0.05). Compared with the true-gating group, the virtual ECG-gating group had a 20.5% higher effective dose [3.99 (3.77, 4.28) mSv vs. 3.31 (2.91, 3.75) mSv, P<0.001], a 24.7% shorter total examination time [(3.62±0.34) min vs. (4.81±0.54) min, P<0.05], and a 61% shorter positioning time [(0.76±0.19) min vs. (1.95±0.36) min, P<0.05], but a 14.6% longer image reconstruction time [(48.06±7.18) s vs. (41.93±8.90) s, P<0.05]. In the ICA-validated subgroup, diagnostic concordance rates were 79.17% in the true ECG-gating group and 92.86% in the virtual ECG-gating group; the difference in diagnostic performance was not statistically significant (AUC: 0.792 vs. 0.854, P>0.05). Conclusions Virtual ECG-gated CCTA significantly improves efficiency while maintaining image quality and diagnostic performance, with a slight increase in radiation dose. It is recommended for time-sensitive clinical scenarios, with adherence to dose-optimization principles.
Objective To evaluate the utility of a mammography-based radiomics nomogram model in predicting human epidermal growth factor receptor 2 (HER-2) expression status in breast cancer. Methods A total of 288 female breast cancer patients who underwent mammography and had surgical pathology confirmation were retrospectively collected from two hospitals. Patients from one hospital (n=215) were randomly divided into a training cohort (n=150) and a validation cohort (n=65) at a 7∶3 ratio; patients from the other hospital (n=73) served as an external test cohort. According to postoperative pathology, all patients were categorized into three groups: HER-2-positive (n=106), HER-2-low (n=128), and HER-2-zero (n=54). Radiomics features were extracted from mammograms, dimension-reduced and screened to construct a radiomics score (Radscore). Laboratory results, imaging T stage, maximum tumor diameter, shape, margin, and density were analyzed to identify clinical and mammographic indicators with statistically significant differences among three groups. Multivariate logistic regression was used to determine independent predictors of different HER-2 expression states, upon which a clinical model and a mammographic-feature model were built; these were then combined with the Radscore to develop nomogram models. Receiver operating characteristic (ROC) curves were used to assess predictive performance. The DeLong test compared differences in AUCs among three models. Decision curve analysis (DCA) evaluated clinical utility. Results For HER-2-positive type, a total of 8 optimal radiomic features were selected to establish Radscore 1. For HER-2-low expression type, 6 optimal radiomic features were selected to establish Radscore 2. Tumor shape and CA153 were independent predictive factors for HER-2-positive，and they were combined with Radscore 1 to construct Nomogram Model 1. Maximum tumor diameter was an independent predictive factor for HER-2 -low expression, and it was combined with Radscore 2 to construct Nomogram Model 2. Nomogram Model 1 achieved the highest AUCs in both the validation and external test cohorts for predicting HER2-positive (DeLong tests, both P<0.001). For predicting HER2-low expression, Nomogram Model 2 showed higher AUC, sensitivity, and specificity than the mammographic-feature Model 2 in both the validation and external test cohorts (DeLong tests, both P<0.001). DCA indicated greater net benefit for both Nomogram Models 1 and 2. Conclusion The radiomics-based nomogram effectively predicts HER-2 expression status in breast cancer, providing important clinical value for guiding HER-2 targeted treatment strategies.
Pancreatic cancer (PC) is one of the most malignant tumors of the digestive system. MRI radiomics can quantitatively analyze tumor heterogeneity features, providing a new perspective for precision diagnosis and treatment of PC. Machine learning-based MRI radiomics has been applied in various aspects of PC management, including differential diagnosis, pathological grading, lymph node metastasis detection, prognosis prediction, immune microenvironment assessment, and predictions of molecular biological expression and gene mutations. These applications can assist clinicians in formulating personalized treatment strategies. This article presents a review of recent progress in the use of machine learning-based MRI radiomics in the diagnosis and treatment of PC.
Pancreatic cancer is a highly malignant tumor of digestive system. Radiomics enables non-invasive, comprehensive, and quantitative analysis of tumor heterogeneity in pancreatic cancer, thereby improving lesion prediction and diagnostic accuracy. Deep learning, through the automatic extraction of deep features from images, further enhances the effectiveness of images in diagnosis and prediction. This review summarizes recent advances in the application of radiomics and deep learning techniques to pancreatic cancer, covering early screening and diagnosis, preoperative staging and metastasis assessment, differential diagnosis, treatment response evaluation, and recurrence and prognosis prediction.
Metal implants have been widely used in various surgical procedures. These metal implants generate artifacts during CT examinations, which can compromise the assessment of surrounding tissues such as muscles and bones. In clinical practice, various reconstruction and post-processing techniques are commonly employed to reduce the impact of metal artifacts on lesion evaluation. In recent years, photon-counting computed tomography (PCCT), with its technical characteristics such as high spatial resolution and multi-energy bin capability, has demonstrated distinct advantages in metal artifact reduction. This paper briefly outlines the principles and advantages of PCCT, introduces the main metal artifact reduction techniques currently available, and reviews the recent progress of its clinical applications in the musculoskeletal system, cardiovascular and cerebrovascular system, head and neck, and nervous system.
Objective To investigate the impact of examination sequence of CT enterography (CTE) and colonoscopy on the incidence of adverse reactions to iodinated contrast media in patients with suspected inflammatory bowel disease (IBD). Methods A total of 485 patients with suspected IBD were prospectively enrolled and randomly assigned, using a random number table, into two groups: Group A (265 cases) underwent colonoscopy followed by CTE, while Group B (220 cases) underwent CTE prior to colonoscopy. The incidence of iodinated contrast media-related adverse reactions was compared between the two groups using the Chi-square test. Results In Group A, a total of 39 patients (14.7%) experienced contrast-related adverse reactions, including 34 cases (12.8%) with mild reactions and 5 cases (1.9%) with moderate reactions. In Group B, two patients (0.9%) experienced only mild adverse reactions. The overall incidence of adverse reactions was significantly higher in Group A than in Group B (χ2=36.82, P<0.001). Conclusion In patients suspected of IBD, performing CTE before colonoscopy can significantly reduce the risk of adverse reactions to iodinated contrast media.
Objective To explore the imaging features of pediatric neuroglial heterotopia in the buccal region. Methods The clinical and preoperative imaging data of two children with pathologically confirmed neuroglial heterotopia in the buccal region were retrospectively analyzed, and relevant literature was reviewed. Results Both children exhibited asymmetry in the cheeks. The lesions were located within the subcutaneous fat layer of the face and presented as lobulated solid or cystic-solid masses with well-defined margins. On CT, the lesions showed soft tissue density, slightly lower than brain gray matter, with relatively uniform density. On MRI, the lesions demonstrated isointense or slight hypointense on T1WI, and slightly hyperintense or isointense mixed signals on T2WI. One case showed mild heterogeneous enhancement, and the other showed peripheral enhancement. Conclusion Facial asymmetry can be detected early in children with neuroglial heterotopia in the buccal region. These lesions present as solid or cystic-solid masses with well-defined margins and density or signal intensity similar to that of brain gray matter. They are often accompanied by abnormal development of the subcutaneous fat layer on the affected side of the cheek.
