Due to its irregular shape and varying contour, pancreas segmentation is a recognized challenge in medical image segmentation. Convolutional neural network (CNN) and Transformer-based networks perform well but have limitations: CNN have constrained receptive fields, and Transformer underutilize image features. This work proposes an improved pancreas segmentation method by combining CNN and Transformer. Point-wise separable convolution was introduced in a stage-wise encoder to extract more features with fewer parameters. A densely connected ensemble decoder enabled multi-scale feature fusion, addressing the structural constraints of skip connections. Consistency terms and contrastive loss were integrated into deep supervision to ensure model accuracy. Extensive experiments on the Changhai and National Institute of Health (NIH) pancreas datasets achieved the highest Dice similarity coefficient (DSC) values of 76.32% and 86.78%, with superiority in other metrics. Ablation studies validated each component’s contributions to performance and parameter reduction. Results demonstrate that the proposed loss function smooths training and optimizes performance. Overall, the method outperforms other advanced methods, enhances pancreas segmentation performance, supports physician diagnosis, and provides a reliable reference for future research.
Objective To investigate the value of contrast-enhanced ultrasonography in detection and diagnosis of small primary liver cancer. Methods SonoVue-enhanced ultrasonography were performed on 353 patients with 378 primary liver cancer, less than 3 cm in diameter. Enhancement patterns and enhancement phases of hepatic lesions on contrast-enhanced ultrasonography were analyzed and compared with the results of histopathology. Results In all hepatic tumors, 96.6% （365/378） lesions enhanced in the arterial phase. Among them, 317 (83.9%) tumors enhanced earlier than liver parenchyma and 48 (12.7%) tumors enhanced synchronously with liver parenchyma, and 342 (90.5%) tumors showed early wash-out in the portal and late phases. With regard to the enhancement pattern, 329 (87.0%) tumors presented whole-lesion enhancement, 35 (9.3%) to be mosaic enhancement and 14 (3.7%) to be rim-like enhancement. If taking the whole-lesion enhancement and mosaic enhancement in arterial phase as diagnotic standard for primary liver cancer on contrast-enhanced ultrasonography, the sensitivity was 92.9%(351/378), and if the earlier or synchronous enhancement of the tumor compared with liver parenchyma in arterial phase and the wash-out in portal phase were regarded as the stardand, the sensitivity was 87.3%(330/378). Conclusion Contrast-enhanced ultrasonography could display real-time enhancement patterns as well as the wash-out processes both in hepatic tumors and the liver parenchyma. It might be of clinical value in diagnosis of primary liver cancer based on the hemodynamics of hepatic tumors on contrast-enhanced ultrasonography.
ObjectiveTo evaluate the diagnostic value of contrast enhanced ultrasound (CEUS) to the sentinel lymph node (SLN) of breast cancer. MethodsSeventy-two operable breast cancer patients with clinically negative axillary lymph node were enrolled.Sulphur hexafluoride microbubbles for injection (SonoVue) was used alone as the tracer agent for the sentinel lymph node biopsy (SLNB), and axillary dissection was performed after the methylene blue location.All SLNs were examined pathologically with HE staining.The SLN diagnosis result of contrast enhanced ultrasound and postoperative pathological examination result were comparative analyzed. ResultsAfter the injection of SonoVue can obtain a clear image of the lymphatic vessels and SLN.The success rate of CEUS imaging was 84.72% (61/72) in this group of 72 patients, and the false negative rate was 12.12% (4/33).The sensitivity and specificity of diagnosis by CEUS was 92.50% (37/40) and 92.59% (50/54), respectively, the diagnostic odds ratio (DOR) was 154.17.By the pathology results as the gold standard, the internal consistency of these two methods was good (Kappa value=0.848, P < 0.01). ConclusionCEUS may be a useful orientation and determination method for SLNs.
Objective To observe the protective effect of ultrasound microbubble contrast agentmediated transfection of brain-derived neurotrophic factor(BDNF) into the retina and visual cortex on retinal ganglion cells (RGC) after optic nerve injury. Methods A total of 88 male Sprague-Dawley (SD) rats were randomly divided into normal group (group A, eight rats), sham operation group (group B, 16 rats), control group (group C, 16 rats), eyes transfection group (group D, 16 rats), brain transfection group (group E, 16 rats), combined transfection group (group F, 16 rats). The optic nerve crush injury was induced, and then the groups B to F were divided into one-week and two-week after optic nerve injury subgroup with eight rats each, respectively. The rats in group B and C underwent intravitreal and visual cortex injection with phosphate buffered solution respectively. The rats in group D and E underwent intravitreal and visual cortex injection with the mixture solution of microbubbles and BDNF plasmids respectively. The rats in group F underwent both intravitreal and visual cortex injection with the mixture solution of microbubbles and BDNF plasmids at the same time. The ultrasound exposure was performed on the rats in group D to F after injection with the mixture solution of microbubbles and BDNF plasmids. One and two weeks after optic nerve injury, RGC were retrogradely labeled with Fluorogold; active caspase-3 protein was observed by immunohistochemistry and the N95 amplitude was detected by pattern electroretinogram (PERG). Results Golden fluorescence can be observed exactly in labeled RGC in all groups，the difference of the number of RGC between the six groups and ten subgroups were significant（F=256.30，65.18；P＜0.01）. Active caspase-3 in ganglion cell layer was detected in group C to F, but not in group A and B. The difference of the N95 amplitude between the six groups and ten subgroups were significant（F=121.56，82.38；P＜0.01）.Conclusion Ultrasound microbubble contrast agent-mediated BDNF transfection to the rat retina and visual cortex can inhibit the RGC apoptosis after optic nerve injury and protect the visual function.
Objective To determine the correlation between central macular thickness (CMT) and the visual function in patients iwht macular edema (ME). Methods The clinical data of 42 eyes of 40 patients with ME which were examined by optical coherence tomography (OCT) and microperimetry (MP-1) were retrospectively analyzed. In 40 patients (42 eyes), diabetic ME (DME) was in 27 eyes，branch retinal vein occlusion was in 11eyes, and central retinal vein occlusion was in 4 eyes. All of the eyes had undergone OCT，MP-1 and best-corrected visual acuity (BCVA) test. Central macular thickness (CMT) was measured by fast macular scans using OCT. Retinal sensitivity (MS) and fixation patterns were evaluated by Mp-1. The position was chosen :2 disc diameters (DD) temporal to the disc and one third of a DD inferior to the centre of the disc. Results The correlation between CMT and BCVA is not significant (r＝－0.429，P=0.069) as well as the correlation between CMT and MS (r＝－0.433，P=0.058). The difference of CMT between the unstable and stable group was significant (F=3.262, P=0.039）. The difference of CMT between the central fixation group and preferred retinal locus (PRL) group was significant (F=3.173, P=0.044). Conclusions BCVA and MS have no significant correlation with CMT. When CMT increases, the fixation stability decreases, fixation location.changes, and PRL occurs.
ObjectiveTo explore the clinical value of low-dose contrast agnet in CT angiography (CTA) for head and neck by SOMATOM Definition Flash CT.MethodsSixty consecutive patients with head and neck vessel diseases examined by CTA in the head and neck were chosen from West China Hospital of Sichuan University from March to July 2015, and then were randomly divided into two groups (the experimental group: n=30, 30 mL contrast medium; the control group: n=30, 50 mL contrast medium). Imaging post processing techniques included curved plannar reconstruction, volume rendering, and maximal intensity projection. CT values of the different level of carotid arteries (aortic arch, carotid bifuracation, and M1 segment of middle cerebral artery) were measured. The artifact of the remaining contract in the jugular vein and overall quality of the image were observed by two senior doctors using double blind method.ResultsAll the patients in the two groups completed CTA for head and neck successfully. The image qualities of the two groups satisfid clinical diagnostic requirements, and there was no difference in the image qualities between the two groups (P>0.05). The evaluation of venous pollution in the experimental group was lighter than that in the control group (P<0.05). The CT values of aortic arch, carotid bifuracation, and M1 segment of middle cerebral artery in the experimental group [(341.3±89.5), (391.0±103.7), (305.0±62.0) HU] were slightly lower than those in the control group [(437.3±83.7), (532.5±113.3), (396.6±93.1) HU], which were statistically significant (P<0.05).ConclusionLow-dose contrast in CTA for head and neck by SOMATOM Definition Flash CT can satisfy the clinical diagnostic requirements, and reduce the dose of contrast agent and venous pollution, with a good clinical value.
ObjectiveTo analyze systematic reviews (SRs) and meta-analyses on the efficacy of pharmacological interventions in the prevention of contrast-induced acute kidney injury (CI-AKI) after percutaneous coronary intervention (PCI). MethodsWe searched the PubMed, The Cochrane Library, EMbase, CNKI, CBM, WanFang Data, and VIP databases to identify SRs and meta-analyses on the pharmacological interventions aimed at preventing CI-AKI after PCI from inception of the databases to February 2022. Two researchers independently screened the literature and extracted data. The PRISMA statement, AMSTAR 2 tool, and GRADE system were adopted to evaluate the reporting quality, methodological quality, and evidence quality of the findings of the included SRs and meta-analyses. ResultsA total of 14 SRs and meta-analyses were identified. Among the drugs used were statins, N-acetylcysteine, sodium bicarbonate, Dan-hong injection, vitamins, salvianolate, adenosine antagonists, nicorandil, allopurinol, alprostadil, furosemide, trimetazidine, probucol, and brain natriuretic peptide. Most of the studies had relatively comprehensive coverage of the items, with the PRISMA scores ranging from 18 to 24.5. The main issues of reporting quality were protocol and registration, search, and funding. The methodological quality of the SRs and meta-analyses was generally low. Items 2 and 7 were the key items with poor scores, and the non-critical items with poor scores were items 3, 10, and 16. All drugs, except furosemide, decreased the incidence of CI-AKI. The quality of evidence ranged from medium to very low according to GRADE. Conclusion Most of the single drugs or drugs combined with hydration show the potential to prevent CI-AKI, however, the overall methodological quality of the included studies is relatively low, and the strength of evidence is generally low.
Objective To investigate the enhancing effect of ultrasound microbubbles on transfection of recombinant adenoassociated virus (rAAV) mediated green fluorecent protein (EGFP) gene into retinal ganglion cells (RGC) in vivo.Methods A total of 40 adult Sprague-Dawley (SD) rats were divided into four groups randomly (group A,B,C,D) with 10 rats in each. Group A was the normal control, in which the rats underwent intravitreal injection with 5 mu;l phosphate buffered solution. The rats in group B underwent intravitreal injection with 5 mu;l recombinant adenoassociated virus encoding EGFP gene (rAAV2-EGFP). The rats in group C underwent ultrasound irradiation on eyes right after intravitreal injection with 5 mu;l rAAV2-EGFP; The ultrasound irradiation was performed on the rats in group D right after intravitreal injection with the mixture solution of microbubbles and rAAV2-EGFP ultrasound. After 21 days, RGC were labeled retogradely with fluogold. Seven days after labeling, the retinal flatmounts and frozen sections were made from five rats in each group. Expression of EGFP reporter gene was observed by laser scanning confocal microscope and evaluated via average optical intensity (AOD) and RGC transfection rate. Labeled RGC were counted to evaluate the adverse effects.Results Green fluorescence can be observed exactly in labeled RGC in B，C，and D groups. The AOD and transfection rate in group D was （95.02plusmn;7.25）% and（20.10plusmn;0.74）% , respectively; which were higher than those in group B and C (F=25.970，25.799；P＜0.01). The difference of the number of RGC among the four groups was not significant(F=0.877,P＞0.05). Conclusion Under the condition of low frequency and with certain energy, ultrasoundmediated microbubble destruction can effectively and safely enhance rAAV delivery to RGC in rats.
Retrieving keyframes most relevant to text from small intestine videos with given labels can efficiently and accurately locate pathological regions. However, training directly on raw video data is extremely slow, while learning visual representations from image-text datasets leads to computational inconsistency. To tackle this challenge, a small bowel video keyframe retrieval based on multi-modal contrastive learning (KRCL) is proposed. This framework fully utilizes textual information from video category labels to learn video features closely related to text, while modeling temporal information within a pretrained image-text model. It transfers knowledge learned from image-text multimodal models to the video domain, enabling interaction among medical videos, images, and text data. Experimental results on the hyper-spectral and Kvasir dataset for gastrointestinal disease detection (Hyper-Kvasir) and the Microsoft Research video-to-text (MSR-VTT) retrieval dataset demonstrate the effectiveness and robustness of KRCL, with the proposed method achieving state-of-the-art performance across nearly all evaluation metrics.
Objective To investigate the value of contrast enhanced ultrasound (CEUS) in evaluating the short term therapeutic response to radiofrequency ablation (RFA) of primary hepatocellular carcinoma. Methods One hundred and ten lesions were studied in 96 patients. Each patient underwent CEUS within a week before RFA, the number, size, border, inner echo and perfusion pattern of lesions were observed. One month after ultrasound-guided RFA, color Doppler flow imaging, CEUS and contrast enhanced computed tomography (CECT, reference standard) were performed to assess the therapeutic response. Results Before RFA, in 96 cases with 110 lesions, 83 lesions showed homogeneous hyper-enhancement and the other 27 heterogeneous hyper-enhancement in arterial phase, and 98 lesions were hypo-enhanced in portal venous phase and late phase and the other 12 iso-enhanced. One month after RFA, 99 of 110 lesions were found no-enhancement in entire CEUS procedure, while 11 lesions showed local enhancement on the edge of lesion. Ninety-six of 110 lesions showed no-enhancement and other 14 with irregular enhancement by CECT. There was no statistical significance between CEUS and CECT (χ2=0.406, Pgt;0.05). Fourteen lesions as tumor residual by CECT were underwent RFA again, and then 1 month after RFA no-enhancement was showed by both CECT and CEUS. Conclusion CEUS can play a role in assessing the short term therapeutic response to RFA of hepatocellular carcinoma.