Objective: This study aimed to evaluate cerebral amide proton transfer signal intensity (SI) among controls, hypoxic–ischemic encephalopathy (HIE) neonates with normal conventional magnetic resonance imaging (HIE/MRI−) findings, and HIE neonates with abnormal conventional MRI (HIE/MRI+) findings.Methods: Forty neonates diagnosed with neonatal HIE and sixteen normal neonates were evaluated. All neonates underwent conventional MRI scans and APT imaging. Cerebral APT SIs were compared to identify cerebral regions with significant APT changes among sixteen controls, thirteen HIE/MRI− patients, and twenty–seven HIE/MRI+ patients.Results: Significantly increased APT SIs were observed in the HIE/MRI− group compared with controls, in the left insula, right occipital lobe, left cingulate gyrus (posterior part), and corpus callosum. Significantly increased APT SIs were found in the HIE/MRI+ group compared with controls, in the right anterior temporal lobe (medial part), anterior parts of the right parahippocampal and ambient gyri, left superior temporal gyrus (middle part), left insula, left cingulate gyrus (posterior part), and right lentiform nucleus. No significant APT SI differences were observed in the cerebellum and brainstem among the three groups.Conclusion: Amide proton transfer imaging plays an important role in detecting hypoxic–ischemic encephalopathy regardless of conventional MRI findings. Changes in APT signal intensity may provide important insights into the characterization of the cerebral internal environment. This study suggests that APT imaging could be used as a complement to conventional MRI in the detection of hypoxic–ischemic encephalopathy in clinical practice.
Exposure to aristolochic acid (AA) is of increased concern due to carcinogenic and nephrotoxic effects, and incidence of aristolochic acid nephropathy (AAN) is increasing. This study characterizes renal alterations during the acute phase of AAN using parametric magnetic resonance imaging (MRI). An AAN and a control group of male Wistar rats received administration of aristolochic acid I (AAI) and polyethylene glycol (PEG), respectively, for six days. Both groups underwent MRI before and 2, 4 and 6 days after AAI or PEG administration. T2 relaxation times and apparent diffusion coefficients (ADCs) were determined for four renal layers. Serum creatinine levels (sCr) and blood urea nitrogen (BUN) were measured. Tubular injury scores (TIS) were evaluated based on histologic findings. Increased T2 values were detected since day 2 in the AAN group, but decreased ADCs and increased sCr levels and BUN were not detected until day 4. Significant linear correlations were observed between T2 of the cortex and the outer stripe of outer medulla and TIS. Our results demonstrate that parametric MRI facilitates early detection of renal injury induced by AAI in a rat model. T2 mapping may be a valuable tool for assessing kidney injury during the acute phase of AAN.
Purpose Accurate breast mass classification is crucial for early breast cancer diagnosis. Deep learning shows promise in computer-aided diagnosis but faces challenges due to limited annotated data and lesion complexity. We propose a novel multi-task transfer learning framework to improve mass classification performance and provide a well-performed framework for medical image analysis. Methods The proposed framework comprises a transfer learning backbone and multi-task-specific branches. Pretraining weights from natural image datasets are leveraged to finetune the backbone network, enhancing the ability to extract breast mass characteristics. The classification branches include the primary task for breast mass classification and auxiliary tasks for BI-RADS evaluation, guiding the model to focus on relevant diagnostic features. Result Three models were compared using ResNet50/InceptionV3 as backbones. The Multi-Task Transfer Learning Framework (MTL + TL) achieved the highest AUC values of 0.852 ± 0.019 / 0.824 ± 0.021, outperforming other models in accuracy (0.7654 ± 0.0218 / 0.7667 ± 0.0224), precision (0.6842 ± 0.0379 / 0.7179 ± 0.0379), and F1-score (0.6842 ± 0.0379 / 0.7179 ± 0.0379). Grad-CAM heatmaps confirmed its effectiveness and ability to capture breast mass characteristics. Conclusion Our proposed framework significantly improved breast mass classification, addressing limited annotated data and providing an effective solution to address the limited availability of annotated data. The proposed framework enhanced feature recognition and overall performance by simulating a clinician's observation and decision-making of imaging features through the auxiliary task. This approach can be extended to other breast lesion classifications and provide valuable guidance for medical imaging analysis.