ABSTRACT
Objective@#To preliminarily analyze the prediction efficiency of multimodal MRI-based radiomics model for preoperative glioma IDH1 gene expression type.@*Methods@#The MRI data of 108 surgery-proven glioma patients from May 2015 to January 2019 were retrospectively analyzed, and the MRI data included axial T1WI,T2WI,fluid attenuated inversion recovery (FLAIR),DWI imaging and enhanced T1WI sequence.Forty-seven cases were IDH1 mutant type, and 61 cases were IDH1 wild type. All patients were divided into training and validation groups according to the 7∶3 ratio of the random forest model. Seventy-three cases were in training group, and 35 cases were in validation group. Independent predictors of IDH1 mutation were screened by univariate analysis combined with multivariate logistic regression analysis (P<0.05) in order to construct a random forest diagnosis model of general clinical information and conventional MRI morphological features.General clinical information and conventional MRI morphological features included gender, age, umbers of cases of left and right hemispheres, location of tumors, maximum diameter of tumors, peritumoral edema, intratumoral cystic degeneration, enhancement and ADC value of tumors. The ROI was manually outlined by MaZda software in the most obvious level of 5 sequences of tumor mass and the radiomics features were extracted, including the gray-level co-occurrence matrix(GLCM), the run-length matrix(RUN), the absolute gradient(GRA),the auto-regressive model(ARM) and wavelets transform(WAV). The least absolute shrinkage and selection operator (LASSO)regression were used to select image radiomics features with a method of 10 fold cross -validation and to reduce the dimensions. The screened image radiomics labels were combined with the conventional morphological feature independent predictors to construct a multimodal MRI-based random forest model, and the validation data set was used to evaluate the accuracy and diagnostic efficiency of each model. The sensitivity and specificity of conventional MRI morphological feature model and multimodal MRI-based radiomics prediction model were evaluated dynamically by drawing ROC curves, and the prediction efficiency of the two models was quantified by using AUC statistical indicators. The model classification error rate under different outcomes and the classification error rate of out of bag(OOB)were used to evaluate the stability of the multimodal MRI-based random forest model. The contribution rate of each variable to the model was reflected by the characteristic variables importance assessment map.@*Results@#Univariate regression analysis of the conventional MRI morphological characteristics showed that peritumoral edema, cystic degeneration and enhancement were the three independent predictors of IDH1 gene expression (P<0.01). LASSO algorithm and 10-fold cross-validation identified six robust radiomic features including high frequency coefficients of wavelet transform (WavEnHH_s-4) of T2WI, S(4,4) inverse difference of gray uniformity measurement (InvDfMom),S(5,0) Entropy (entropy),WavEnHH_s-4 of T1WI enhancement, S(1,1) InvDfMom,S(1, -1) Entropy Difference (DifEntrp)of Flair.The error rate of classification for different outcomes and classification error rate of random forest OOB data of multimodal MRI radiomics diagnosis model finally stabilized at 10%. The results of Characteristic Variable Importance Assessment Map: Mean Decrease Accuracy and Mean Decrease Gini index were consistent, which showed that besides three conventional MRI morphological predictors peritumoral edema, enhancement and cystic degeneration, the radiomics labels also played a key role in the model. The results of ROC curve showed that the accuracy, specificity, sensitivity and AUC of conventional MRI morphological feature model were 82.7%, 68.4%, 90.9% and 0.835, respectively, and those of multimodal MRI-based radiomics model were 88.5%, 89.5%, 87.8% and 0.956 respectively.@*Conclusion@#Multimodal MRI-based radiomics random forest model can improve the predictive efficiency of preoperative glioma IDH1 gene expression type more quantitatively.
ABSTRACT
Objective To preliminarily analyze the prediction efficiency of multimodal MRI?based radiomics model for preoperative glioma IDH1 gene expression type. Methods The MRI data of 108 surgery?proven glioma patients from May 2015 to January 2019 were retrospectively analyzed, and the MRI data included axial T1WI,T2WI,fluid attenuated inversion recovery (FLAIR),DWI imaging and enhanced T1WI sequence.Forty-seven cases were IDH1 mutant type, and 61 cases were IDH1 wild type. All patients were divided into training and validation groups according to the 7∶3 ratio of the random forest model. Seventy-three cases were in training group, and 35 cases were in validation group. Independent predictors of IDH1 mutation were screened by univariate analysis combined with multivariate logistic regression analysis (P<0.05) in order to construct a random forest diagnosis model of general clinical information and conventional MRI morphological features.General clinical information and conventional MRI morphological features included gender, age, umbers of cases of left and right hemispheres, location of tumors, maximum diameter of tumors, peritumoral edema, intratumoral cystic degeneration, enhancement and ADC value of tumors. The ROI was manually outlined by MaZda software in the most obvious level of 5 sequences of tumor mass and the radiomics features were extracted, including the gray?level co?occurrence matrix(GLCM), the run?length matrix(RUN), the absolute gradient(GRA),the auto?regressive model(ARM) and wavelets transform (WAV). The least absolute shrinkage and selection operator (LASSO)regression were used to select image radiomics features with a method of 10 fold cross?validation and to reduce the dimensions. The screened image radiomics labels were combined with the conventional morphological feature independent predictors to construct a multimodal MRI?based random forest model, and the validation data set was used to evaluate the accuracy and diagnostic efficiency of each model. The sensitivity and specificity of conventional MRI morphological feature model and multimodal MRI?based radiomics prediction model were evaluated dynamically by drawing ROC curves, and the prediction efficiency of the two models was quantified by using AUC statistical indicators. The model classification error rate under different outcomes and the classification error rate of out of bag(OOB)were used to evaluate the stability of the multimodal MRI?based random forest model. The contribution rate of each variable to the model was reflected by the characteristic variables importance assessment map. Results Univariate regression analysis of the conventional MRI morphological characteristics showed that peritumoral edema, cystic degeneration and enhancement were the three independent predictors of IDH1 gene expression (P<0.01). LASSO algorithm and 10?fold cross?validation identified six robust radiomic features including high frequency coefficients of wavelet transform (WavEnHH_s?4) of T2WI, S(4, 4) inverse difference of gray uniformity measurement (InvDfMom), S(5, 0) Entropy (entropy), WavEnHH_s?4 of T1WI enhancement, S(1, 1) InvDfMom, S(1, -1) Entropy Difference (DifEntrp)of Flair.The error rate of classification for different outcomes and classification error rate of random forest OOB data of multimodal MRI radiomics diagnosis model finally stabilized at 10%. The results of Characteristic Variable Importance Assessment Map: Mean Decrease Accuracy and Mean Decrease Gini index were consistent, which showed that besides three conventional MRI morphological predictors peritumoral edema, enhancement and cystic degeneration, the radiomics labels also played a key role in the model. The results of ROC curve showed that the accuracy, specificity,sensitivity and AUC of conventional MRI morphological feature model were 82.7%, 68.4%, 90.9% and 0.835, respectively, and those of multimodal MRI?based radiomics model were 88.5%, 89.5%, 87.8% and 0.956 respectively. Conclusion Multimodal MRI?based radiomics random forest model can improve the predictive efficiency of preoperative glioma IDH1 gene expression type more quantitatively.
ABSTRACT
Objective To investigate the feasibility of low-c oncentration iso_osmolar contrast agent together with low tube voltage and iterative reconstruction algorithm in rabbit liver computed tonography (CT) perfusion imaging.Methods A total of 15 bealthy New Zealand rabbits were scanned twice of liver CT perfusion scans each with 24 hours interval.The first scan (routine group) was acquired at 100 kV and 100 mAs with ultravist (370 mg/ml),while the second (double low group) was acquired at 80 kV and 100 mAs with iodixanol (270 mg/ml) at 24 hours after the first scan.The obtained images were reconstructed with filtered back projection (FBP) and adaptive iterative dose reduction (AIDR-3D)algorithms in the controlled and experimental groups,respectively.The perfusion parameters including hepatic artery perfusion(HAP),portal vein perfasion(PVP),hepatic perfusion index(HPI),and total liver perfusion(TLP) and image quality as image quality score,average CT value of abdomen aorta,signalto-noise ratio(SNR),carrier-to-noise ratio(CNR),and figure of merit(FOM) were compared used pair ttest or Mann-Whitney U-test between the two groups wherever appropriate.The effective radiation dose and iodine intake were also recorded and compared.Results The image quality and perfusion parameters had no significantly different between the two groups except for FOM.The effective radiation dose and iodine intake were 38.79％ and 27.03％ lower in the double low group.Conclusions Low concentration iso _osmolar contrast agent (iodixanol,270 mg/ml) together with low tube voltage (80 kV) helps to reduce radiation dose and iodine intake without compromising perfusion parameters and image quality in liver CT perfusion imaging.
ABSTRACT
Objective To explore the feasibility of coronary computed tomographic angiography (CCTA) for obese patients with lower tube voltage (100 kV) and lower contrast media concentration (270 mgI/ml) using iterative reconstruction.Methods A total of 48 patients with body mass index greater than 30 kg/m2 were included and randomly divided into 2 groups according to random number table method.The images of the control group were obtained using iodine 370 mgI/ml, a tube voltage of 120 kV, and traditional filtered back projection (FBP) image reconstruction.Patients in the test group were injected with isotonic low concentration contrast media (270 mgI/ml), scanned with a lower tube voltage (100 kV), and adaptive iterative noise reduction image reconstruction algorithm (AIDR-3D) was used.Two experienced physicians scored the image quality in a double-blind way.Independent sample t-test was used to compare the effective dose (E), average CT values, signal to noise ratio (SNR), contrast to noise ratio (CNR), the figure of merit (FOM), image quality scores and the total iodine intake.Side effect was also evaluated.Results The subjective scores for control group and test group were not significantly different (P ＞ 0.05).The scores of two physicians were consistency (Kappa =0.88, P ＜ 0.05).The average CT values, SNR and CNR for the two groups were not significantly different (P ＞ 0.05), but the FOM of the test group was significantly higher than that of the control group (t =-9.250,-8.604,-9.158,-5.341, P ＜ 0.05).Effective dose in the test group was (1.61 ± 0.41) mSv, lower than that of the control group (t =8.373, P ＜ 0.01).The total iodine and iodine injection rate in the test group were both lower than in the control group (t =7.628, 8.480, P ＜ 0.01).The incidence of contrast mediarelated discomfort in the test group was lower than control group (x2 =18.70, 6.25, P ＜ 0.05).Conclusions For obese patients, isotonic low concentration of contrast media and low-dose CCTA could be feasible, which substantially reduce the radiation dose and iodine intake without sacrificing image quality.Trial registration Chinese clinical trial registry, ChiCTR-DPD-15007510.
ABSTRACT
Objective To compare the clinical diagnostic value of 320 row four-dimensional CT angiography (4D-CTA)and digital subtraction angiography (DSA)in untreated arteriovenous malformation (AVM)patients.Methods Thirty-six patients with AVM diagnosed by DSA were included in this study.Two independent readers blind to the results of all examinations evaluated the find-ings of DSA and 4D-CTA of each patient.All results were then documented using a standardized scoring sheet.Results The diagno-sis results of 4D-CTA for AVM lesions in all 36 patients were accurately consistent with those of DSA,including the position,size and vascular structure.There were 14 cases with the largest diameter 6 cm.Lesions involved only the anterior circulation in 13 cases,only the posterior circulation in 13 cases,and both the anterior and posterior circulation in 10 cases.Conclusion 4D CTA has excellent diagnostic accuracy in the detection of AVM lesions,including size,location,feeding ar-teries and draining veins,with similar value as DSA in the clinical diagnose and evaluation of AVM.
ABSTRACT
Objective To evaluate the image quality and radiation dose of prospective electrocardiography-triggered coronary 320-slice volume CT angiography with different kV, and the feasibility of coronary scan with ＜ 1 mSv radiation dose.Methods Eighty consecutive patients were randomly divided into two groups equally.The tube voltage according to paradigm was 100 kV in group A and 120 kV in group B.All raw data in group A was reconstructed by the software AIDR in CT system to create a new group named as A1. Such parameters as the mean intraluminal attenuation (SI),noise (SD),signal-to-noise ratio (SNR),contrast-to-noise ratio (CNR),effective radiation dose(E) and image quality score measured in group A were compared with those in group B.The values such as SI,SD,SNR,CNR,image quality scores were compared between group A and group A1.The significance of group B and group A1 was compared in SI,SD,SNR,CNR,image quality scores as well.Results E in group A was significantly lower than that in group B[ E =(0.67 ± 0.18) mSv in group A vs.E =(3.08 ± 1.04) mSv in group B].The value of E in group A was decreased by 78％ compared to group B(t =- 14.30,P＜0.05 ).There was no significant difference in mean image quality scores between two groups(4.57 ± 0.57in groupA vs.4.59 ± 0.59 in group B,t=-1.17,P＞0.05).The values of SI,SD,SNR,CNR in group A were (570.8 ±131.5)HU,25.1 ±6.9,24.5 ±9.1,19.8 ±6.1.And the values of SI,SD,SNR,CNR in group B were (460.6 ± 14.3) HU,15.1 ±3.6,31.7 ±7.7,29.3 ±6.8.The values of SI and SD in group A were significantly higher than those in group B(t =4.49,8.18,P ＜0.05). The values of SNR and CNR in group A were lower than those in group B (t =-4.24,-6.19,P＜0.05).The valuesofS1,SD,SNR,CNR,image quality scores in group Al were (557.9 ±24.5) HU,21.1 ±6.0,27.7±10.0,23.4±7.8,4.60 ± 0.56.There was no difference in the SI and the image quality scores between group A and group A1 ( t =1.09,- 1.90,P ＞ 0.05).Conclusions 320-slice volume CT with 100 kV tube voltage and prospective ECG-triggered technique can reduce the radiation dose to less than 1 mSv and obtain optimal images in diagnosis of coronary arterial diseases.
ABSTRACT
Objective To evaluate perfusion computed tomography in the assessment of portal vein pressure changes in an experimental dog model of liver cirrhosis and portal hypertension.Methods The canine model of cirrhosis and portal hypertension was induced by portal vein stenosis with combination of systemic thioacetamide(TAA) feeding in drinking water.All of the Beagles in control group and cirrhotic group underwent hepatic perfusion on a spiral CT scanner.The parameters of hepatic perfusion were calculated by the method of deconvolution.The portal vein pressure was measured by a laparotomy surgery.Results ① In control group, the portal vein pressure was ( 14.5 ± 2.2) cm H2O, while it was (23.1 ± 2.8) cm H2O in PHT group, there was significant difference in the portal vein pressure between the two groups (P＜0.05).② The blood flow(BF) was (112 ±14) ml·100 g-1·min-1 in controls, while ( 96 ± 11) ml·100 g-1·min-1 in PHT group; the blood volume ( BV ) in control group and PHT group was (10 ±3) ml·100 g-1 and (11 ± 5) ml· 100 g-1, respectively; the mean transit time( MTT) was (7.1 ± 2.0) s and (10.4 ± 3.5) s, respectively; the hepatic arterial fraction (HAF) was ( 24 ± 5) % and ( 37 ± 6)% , respectively; the hepatic arterial perfusion (HAP) was(27 ±6) ml·100 g-1·min-1 and (35 ±5) ml·100 g-1·min-1, respectively; the portal venous perfusion (PVP) was (85 ± 13) ml·100 g-1·min-1 and (61 ±11) ml·100 g-1·min-1, respectively.There was significant difference in all parameters between the two groups except the parameter BV(P ＜ 0.05).③ In PHT group, the PVP and BF were negatively correlated with the portal vein pressure, while positively correlated with MTT and HAF.Portal vein pressure was negatively correlated with PVP, the equation, Y = 36.624 -0.219X, was deduced with linear regression analysis, by which the portal vein pressure in PHT Beagles was ( 23.2 ± 2.4) cm H2O, which was correlated with the observed by laparotomy value (23.1 ± 2.8) cm H2O (r = 0.843, P ＜ 0.05).Conclusion CT perfusion is a new non-invasive and effective method for assessment of portal vein pressure.
ABSTRACT
Objective To investigate the feasibility and efficacy of iliofemoral endovascular therapy for the treatment of chronic ischemia of lower extremities due to arterial atherosclerosis. Methods During the period of September 2005-January 2009 percutaneous endovascular angioplasty and stent implantation were performed in 15 consecutive patients with CTA-proved chronic occlusive iliofemoral artery disease. The patients included 11 males and 4 females with a mean age of (72.1 ± 5.9) years (ranged from 61 to 82 years). Before the procedure, all patients underwent clinical and imaging assessments, including Fontaine classification, ankle-brachial pressure index (ABI), and lower limb computed tomography angiography (CTA). According to the length, morphology and location of the occlusive segment, different types of stents were employed together with pereutaneous transluminal angioplasty (PTA) to treat the occluded lilac or superficial femoral artery. At the end of the procedure, distal angiogram was performed to assess the success of the procedure and to exclude thromboembolism or dissection complications. After PTA, the residual stenosis < 30% and/or pressure gradient < 10 mmHg were defined as technical success. Clinical success was identified when patient's clinical symptoms were improved by one or more grade according to Fontaine classification. Results Technical success was totally achieved in all patients with no occurrence of complication. Recanalization of the occluded artery was performed preferably by retrograde ipsilateral femoral artery approach in 5 patients and by contralateral access using the crossover aorta technique in 10 patients. PTA was performed in all patients after stent release to open the occluded vessel to its normal diameter. The mean length of the lilac arterial segment covered with 8 stents was (7.6 ± 1.4) cm (ranged from 6 cm to 10 cm). The mean diameter ± SD of the iliac stents was 9.5 mm ± 1.4 mm (ranged from 8 mm to 12 mm). The mean length of 5 superficial artery implanted stents was (4.5 ± 1.1) cm (ranged from 4 cm to 6 cm). The mean ABI before treatment and six months after treatment was 0.64 (range 0.41 to 0.89) and 0.76 (range 0.50 to 0.95), respectively. The difference in ABI between pre-interventional and post-interventional procedure was statistically significant (t = -4.64, P < 0.01). Clinical improvements according to Fontaine classification were obtained in all patients, in whom technical success was achieved (100%). Conclusion Endovascular interventional procedure is a safe and effective treatment for iliofemoral arterial occlusive disease due to arterial atherosclerosis. PTA combined with subsequent stent implantation can well improve the patient's quality of life as well as the hemodynamics of lower extremities, which is very important for preserving the diseased limb for a long period of time.