Objective To explore the value of curvature value of liver surface nodularity(LSN)based on MRI in evaluating liver function in patients with liver cirrhosis.Methods A retrospective analysis was made on the patients who underwent upper abdomen MR examination at 3.0T.The normal liver function patients and cirrhosis patients were enrolled in the study and then the Child-Pugh score of the patients were calculated.The patients were divided into three groups:normal liver group,compensated cirrhosis group and decompensated cirrhosis group.The water phase imaging of 3D modified Dixon fast field echo(mDixon-FFE)sequence was copied in DICOM format.ITK software was used to manually draw the full-thickness liver edge by two observers.The curvature value of LSN was obtained by using matlab self compiled code for follow up analysis.Kruskal-Wallis H test was used to compare the curvature value between the groups.The receiver operating characteristic(ROC)curve was drawn and the area under the curve(AUC)was obtained.Spearman test was used for the correlation analysis.Results The curvature values of LSN among the normal liver,compensated cirrhosis and decompensated cirrhosis groups gradually increased(P<0.05).Comparing normal liver with compensated cirrhosis,the AUC of diagnosing compensated cirrhosis was 0.84,with the sensitivity of 72.7%and the specificity of 89.3%.Comparing compensated cirrhosis with decompensated cirrhosis,the AUC of diagnosing decompensated cirrhosis was 0.91,with the sensitivity of 80%and the specificity of 90.9%.There was a moderate positive correlation between the curvature value of LSN and liver function score in patients with cirrhosis(r=0.63,P=0.002).Conclusion The curvature value of LSN based on MRI can be used for preliminary evaluation of liver function of liver cirrhosis,with the AUC more than 0.80 and higher sensitivity and specificity.

Objective:To investigate the association between air pollutant exposure levels and the risk of atrial fibrillation and to evaluate the mediating role of serum uric acid levels in the association between air pollutant levels and atrial fibrillation risks.Methods:This study was a case-control study, and the data of the atrial fibrillation group was derived from atrial fibrillation patients diagnosed at the Zhongda Hospital, Affiliated to Southeast University, from January 2014 to April 2021, and data of control group was derived from those without atrial fibrillation at the screening in Qixia District, Nanjing City, in March to April 2019. A 1∶1 propensity score matching was performed with the matched variables of age and sex. Air pollutant exposure data were collected from 9 air quality monitoring stations in Nanjing from January 2014 to April 2021, including NO 2, CO, PM 2.5 and PM 10. Exposure to air pollutants was converted to respiratory exposure levels according to an approximate formula. Multivariate logistic regression models were used to analyze the association between air pollutants and the risk of atrial fibrillation. Mediation analysis was used to investigate the mediating role and magnitude of serum uric acid in the association pathway between the four air pollutants (PM 2.5, PM 10, NO 2, and CO) and atrial fibrillation. Results:The atrial fibrillation group was aged (68.7±11.3) years, with 544 (51.8%) males; the control group was aged (68.5±8.9) years, with 543 (51.7%) males. Multivariate logistic regression models suggested that individual exposure levels of all four air pollutants were associated with the increased risks of atrial fibrillation. Every 1 μg·kg -1·d -1 increased in NO 2 was associated 12.1% increased risk of atrial fibrillation among the individuals ( OR=1.121, 95% CI:1.098-1.144); For every 1 μg·kg -1·d -1 increased in CO, the individual risk of atrial fibrillation increased by 0.7% ( OR=1.007, 95% CI: 1.006-1.008). For each 1 μg·kg -1·d -1 increase in PM 2.5 exposure level, the individual risk of atrial fibrillation increased by 14.2% ( OR=1.142, 95% CI: 1.120-1.164). For each 1 μg·kg -1·d -1 increase in PM 10 exposure level, the individual risk of atrial fibrillation increased by 3.7% ( OR=1.037, 95% CI: 1.028-1.046). The results of the mediation analysis suggested that serum uric acid levels mediated 5.6% ( P=0.032) causal effects of PM 2.5 on atrial fibrillation risks and 7.5% ( P=0.010) mediated by CO. Conclusion:Air pollutant exposure was a risk factor for the development of atrial fibrillation and uric acid mediated the increased risk of atrial fibrillation by PM 2.5 and CO.

Purpose To explore the application value of 1 024×1 024 reconstruction matrix combined with iterative reconstruction algorithm(Karl)in computed tomography angiography(CTA)of gastric vessels and tumor-supplying arteries.Materials and Methods Thirty patients with gastric tumors who underwent clinical gastric CTA from March to June 2022 at the First Affiliated Hospital of Dalian Medical University were prospectively collected.The original data of CT scan images were grouped and reconstructed.Group A used the conventional 512×512 matrix combined with Karl-5 reconstruction;group B used 1 024×1 024 large reconstruction matrix combined with different levels of Karl algorithm reconstruction,obtaining three subgroups:B1(Karl 5),B2(Karl 7)and B3(Karl 9).The CT and SD values of abdominal aorta at the origin of the left gastric artery,celiac trunk,splenic artery,hepatic artery and the subcutaneous fat tissue of the abdominal wall were measured,and the signal-to-noise ratio(SNR)and contrast-to-noise ratio(CNR)were calculated.The image quality of each group was assessed by two observers using a 5-point scale.Results The two observers had good consistency in the evaluation of image quality(Kappa=0.782-0.789,P<0.05).Subjective scores for groups B1 to B3 were all better than those for group A,with the highest score in group B2(two-dimensional scores:χ2=27.309,24.250;three-dimensional scores:χ2=21.964,21.294;all P<0.05).The gastric artery system was involved in tumor blood supply in 25 cases of 30 gastric tumors,with the clarity rate of each vessel in group B2 being superior to group A.There were no statistically significant differences in CT values of vessels between groups A,B1 and B2(t=-1.918-2.720,P>0.05),and apart from the SD of the abdominal aorta and CNR of the hepatic artery,differences in SD values,SNR,CNR,and background noise between group B2 and group A were not significant(t=-5.909-5.768,P>0.05);images from B1 to B3 showed a gradual decrease in SD values(F=2.881-27.109,P<0.05)and a gradual increase in SNR and CNR(F=3.612-12.149,P<0.05)as the level of Karl algorithm increased.Conclusion 1 024×1 024 reconstruction matrix combined with Karl-7 can improve the image quality and enhance the display effect of gastric microvessels and tumor-supplying artery branches,which have good clinical application value.

Purpose To study the changes of white matter microstructure in moderate and severe obstructive sleep apnea-hypopnea syndrome(OSAHS)via synthetic MRI.Materials and Methods A total of 30 patients with moderate and severe OSAHS diagnosed for the first time and 30 healthy controls were prospectively collected in General Hospital of Taiyuan Iron&Steel(Group)Co.,LTD from October 2022 to September 2023.Both groups underwent routine MRI,3D-T1 MPRAGE and synthetic MRI scans.The T1,T2 and proton density(PD)values of multiple white matter regions in the two groups were compared,and the correlation between T1,T2 and PD values in different brain regions and sleep parameters was analyzed.Results Compared with the control group,the T1 value were increased in various white matter regions,including the right cerebral peduncle,cingulate fasciculus,posterior thalamic radiation,external capsule,bilateral internal capsule,corona radiata,superior longitudinal fasciculus and superior frontal-occipital fasciculus(t/Z=-3.267-3.286,all P<0.05).The T2 value of the left inferior cerebellar peduncle was decreased(Z=2.114,P=0.035),and the T2 value of the right posterior thalamic radiation and bilateral external capsule were increased(Z=-2.410,-2.484,-2.306,all P<0.05).Furthermore,the PD value were increased in corpus callosum,left internal capsule,sagittal layer,hippocampus,right posterior thalamic radiation,cingulate fasciculus,bilateral corona radiata and superior longitudinal fasciculus(t/Z=-3.726-2.757,all P<0.05).The apnea hypopnea index of patients with moderate and severe OSAHS was positively correlated with the T1 value(r=0.387,P=0.035)and PD value(r=0.382,P=0.037)of the right cingulate bundle,and there was no correlation between the lowest oxygen saturation and the quantitative parameters in moderate and severe OSAHS(P>0.05).Conclusion Synthetic MRI can sensitively observe the changes of white matter relaxation values in moderate and severe OSAHS,and has high application value in evaluating the damage of white matter in OSAHS.

Objective:To investigate the value of quantitative parameters of amide proton transfer-weighted (APTw) imaging and dynamic contrast-enhanced (DCE)-MRI for preoperative assessment of human epidermal growth factor receptor 2 (Her-2) gene expression in endometrial cancer (EC).Methods:This research conducted a diagnostic pilot study involving 68 patients with pathologically confirmed EC at the First Hospital of Dalian Medical University from August 2019 to August 2023. Patients were categorized into Her-2-positive group (33 cases) and Her-2-negative group (35 cases) based on postoperative Her-2 gene expression results. Utilizing the APTw and DCE-MRI sequences, quantitative parameters including the asymmetric magnetization transfer ratio (MTR asym) for APTw and the volumetric transfer constant (K trans), plasma volume fraction (V p), extracellular mesenchymal space (V e), and rate constant (K ep) for DCE-MRI were acquired for the lesion site. Statistical differences in the values of each quantitative parameter between the two groups were evaluated using two independent sample t test or Mann-Whitney U test. The study incorporated quantitative parameters and clinicopathological data of patients to identify independent predictors of EC Her-2 gene expression through logistic regression analysis. A diagnostic model was developed using binary logistic regression analysis. The effectiveness of the parameters and diagnostic model was evaluated using receiver operating characteristic curves. DeLong test was used to compare the differences between the areas under the curves (AUC). Results:The study found statistically significant differences in MTR asym, K trans, and V e between the Her-2-positive group and the Her-2-negative group ( Z=2.55, P=0.011; t=-2.03, P=0.047; t=-2.13, P=0.037). However, the differences in V p and K ep were not statistically significant ( Z=0.58, P=0.560; Z=0.19, P=0.849). MTR asym emerged as a significant independent predictor of Her-2 gene expression in EC ( OR=1.016, 95% CI 1.003-1.030, P=0.014). Incorporating MTR asym, K trans, and V e, the diagnostic model yielded an AUC (95% CI) of 0.745 (0.625-0.864). The AUC (95% CI) for MTR asym, K trans, and V e alone were 0.680 (0.551-0.808), 0.623 (0.485-0.760), and 0.656 (0.523-0.789) respectively. The differences in AUC between the diagnostic model and individual predictors MTR asym, K trans, and V e were not found to be statistically significant ( Z=1.40, 1.92, 1.37, P=0.163, 0.055, 0.171). Conclusion:The quantitative parameters of APTw and DCE-MRI sequences can preoperatively assess EC Her-2 gene expression from a different perspective, with MTR asym potentially serving as a valuable independent predictor.

Recently, deep learning has achieved impressive results in medical image tasks. However, this method usually requires large-scale annotated data, and medical images are expensive to annotate, so it is a challenge to learn efficiently from the limited annotated data. Currently, the two commonly used methods are transfer learning and self-supervised learning. However, these two methods have been little studied in multimodal medical images, so this study proposes a contrastive learning method for multimodal medical images. The method takes images of different modalities of the same patient as positive samples, which effectively increases the number of positive samples in the training process and helps the model to fully learn the similarities and differences of lesions on images of different modalities, thus improving the model's understanding of medical images and diagnostic accuracy. The commonly used data augmentation methods are not suitable for multimodal images, so this paper proposes a domain adaptive denormalization method to transform the source domain images with the help of statistical information of the target domain. In this study, the method is validated with two different multimodal medical image classification tasks: in the microvascular infiltration recognition task, the method achieves an accuracy of (74.79 ± 0.74)% and an F1 score of (78.37 ± 1.94)%, which are improved as compared with other conventional learning methods; for the brain tumor pathology grading task, the method also achieves significant improvements. The results show that the method achieves good results on multimodal medical images and can provide a reference solution for pre-training multimodal medical images.
Humans ; Algorithms ; Brain/diagnostic imaging* ; Brain Neoplasms/diagnostic imaging* ; Recognition, Psychology

Hepatocellular carcinoma (HCC) is the most common liver malignancy, where HCC segmentation and prediction of the degree of pathological differentiation are two important tasks in surgical treatment and prognosis evaluation. Existing methods usually solve these two problems independently without considering the correlation of the two tasks. In this paper, we propose a multi-task learning model that aims to accomplish the segmentation task and classification task simultaneously. The model consists of a segmentation subnet and a classification subnet. A multi-scale feature fusion method is proposed in the classification subnet to improve the classification accuracy, and a boundary-aware attention is designed in the segmentation subnet to solve the problem of tumor over-segmentation. A dynamic weighted average multi-task loss is used to make the model achieve optimal performance in both tasks simultaneously. The experimental results of this method on 295 HCC patients are superior to other multi-task learning methods, with a Dice similarity coefficient (Dice) of (83.9 ± 0.88)% on the segmentation task, while the average recall is (86.08 ± 0.83)% and an F1 score is (80.05 ± 1.7)% on the classification task. The results show that the multi-task learning method proposed in this paper can perform the classification task and segmentation task well at the same time, which can provide theoretical reference for clinical diagnosis and treatment of HCC patients.
Humans ; Carcinoma, Hepatocellular ; Liver Neoplasms ; Learning

Objective:To explore the consistency of MRI fast field echo resembling a CT using restricted echo-spacing (FRACTURE) and CT in the evaluation of knee and ankle bone changes.Methods:From November 2020 to November 2021, seventeen patients who underwent CT and MRI FRACTURE examinations of knee joint or ankle joint in the First Affiliated Hospital of Dalian Medical University were retrospectively collected, including 14 patients with knee joint examinations and 3 patients with ankle joint examinations. According to the number of joint components, 80 components were included, including 14 for femur and patella, 17 for tibia and fibula, and 3 for talus, scaphoid, medial cuneiform, medial cuneiform, lateral cuneiform and calcaneus, respectively. The fracture, hyperosteogeny, and bone destruction of the joint bones were evaluated by two observers using CT and FRACTURE images, respectively. Kappa test was used to analyze the consistency of CT and FRACTURE images between observers in the evaluation of joint bone lesions.Results:The Kappa values (95%CI) of the consistency evaluation of fracture, hyperosteogeny, and bone destruction by CT and FRACTURE images were 0.925 (0.823-1.027), 0.905 (0.799-1.011) and 0.895(0.752-1.038) respectively for observer 1, and were 0.963 (0.892-1.034), 0.933 (0.843-1.023) and 0.886 (0.731-1.041) respectively for observer 2. The Kappa values (95%CI) of the consistency evaluation of fracture, hyperosteogeny, and bone destruction by observers 1 and 2 via CT images were 1.000 (1.000-1.000), 0.937(0.851-1.023) and 0.945 (0.839-1.051) respectively, and that by FRACTURE images were 0.962 (0.888-1.036), 0.966 (0.899-1.033) and 0.836 (0.656-1.016) respectively.Conclusion:For the evaluation of fracture, hyperosteogeny, and bone destruction of knee joint and ankle joint, MRI FRACTURE sequence is highly consistent with CT.

Objective:To explore the value of echo-planar imaging correction (EPIC) for improving image quality of diffusion weighted imaging (DWI) and diffusion tensor imaging (DTI) of cervical cord.Methods:A total of 33 subjects (20 males, 13 females) were scanned on a 3.0 T MR scanner from January to March 2022, and the sequences included T 1WI, DWI and DTI (with and without corrections). Two observers delineated the regions of interest (ROIs) on the fused images of DWI and DTI with T 1WI before and after correction, and measured the average diffusion coefficient (ADC), fractional anisotropy (FA), and offset distance of ROIs between images with and without corrections. The subjective scores of image quality were also evaluated. The ICC or Kappa was used to test the consistency of the quantitative measurement and subjective scores by the two observers. The average values by the two observers would be used for subsequent analysis. The independent pair t-test and Wilcoxon test were used for comparison of objective measurements and Mann-Whitney U test was used for subjective image assessments between images with and without corrections. Results:The measurement data and the subjective scores of the two observers were in good agreement (ICC 0.912-0.999, Kappa 0.778-0.816). The independent sample t-test showed the subjective scores were significantly different for the DWI and DTI images between before and after geometry and/or ADC corrections. The ADC values of C6, the offset distances measured by DWI before and after correction of C4, C5, and C6 and subjective scores were significantly different ( P<0.05); The FA values of C1 and C3, ADC values of C1 and C3, offset distance of C4, C5 and C6 measured by DTI before and after correction and subjective scores were statistically significant ( P<0.001). Conclusion:EPI geometry correction and ADC value correction can significantly reduce geometric distortion, increase image quality, and thus improve the diagnosis accuracy of essential diseases.

Objective:To explore rational surgical treatment for childhood nail matrix nevi.Methods:A retrospective analysis was conducted on clinical data from 35 children with pathologically confirmed nail matrix nevi, who received surgical treatment in Children′s Hospital of Chongqing Medical University from September 2015 to March 2019. Different surgical approaches were adopted according to the site and width of lesions. For lesions with a width of ≤ 3 mm, the nail bed and nail matrix lesions were directly excised with 1-to-2-mm margins and sutured in 11 cases. For lesions with a width of > 3 mm, one of the following 3 surgical procedures was selected by the children′s parents: (1) shaving of nail bed and nail matrix lesions under a microscope at ×8 magnification (8 cases) ; (2) excision of lesions followed by full-thickness skin grafting on the periosteum of the phalanx (8 cases) ; (3) excision of lesions of the second to fifth fingers followed by transfer of skin flaps from the thenar muscle area and full-thickness skin grafting (5 cases) , or excision of lesions of the thumb followed by abdominal-wall flap transfer (3 cases) . The patients were followed up for 12 months, and clinical efficacy was evaluated.Results:During the follow-up, no recurrence occurred in the 11 cases receiving direct excision and suture, with good appearances and longitudinal linear scars on the nail. Among the 8 cases receiving shaving therapy under a microscope, 4 experienced relapse during the follow-up of 6 - 12 months, and the nail/toenail plates were rough and poor in lustrousness in the other 4 without recurrence. No recurrence was observed in the 8 cases receiving excision of the lesions and full-thickness skin grafting, of whom 1 experienced skin graft necrosis, and skin grafts survived with obvious pigmentation in the other 7 cases. Among cases receiving excision of the lesions combined with transfer of skin flaps from the thenar muscle area or abdominal-wall flap transfer, no recurrence was observed, and all transferred flaps survived; good appearances, nearly normal color and gloss of nails were obtained in the cases after transfer of skin flaps from the thenar muscle area, while the color and gloss of postoperative nails were markedly different from those of normal nails in the cases receiving abdominal-wall flap transfer.Conclusion:For nail matrix nevi with a width of ≤ 3 mm, direct excision and suture with 1-to-2-mm margins are recommended; for those with a width of > 3 mm, excision of lesions combined with full-thickness skin grafting, transfer of skin flaps from the thenar muscle area or abdominal-wall flap transfer is recommended; the shaving procedure under a microscope should be used with caution.