Quantification of Myocardial Blood Flow by Machine Learning Analysis of Modified Dual Bolus MRI Examination.

Contrast-enhanced magnetic resonance imaging (MRI) is a promising method for estimating myocardial blood flow (MBF). However, it is often affected by noise from imaging artefacts, such as dark rim artefact obscuring relevant features. Machine learning enables extracting important features from such noisy data and is increasingly applied in areas where traditional approaches are limited. In this study, we investigate the capacity of machine learning, particularly support vector machines (SVM) and random forests (RF), for estimating MBF from tissue impulse response signal in an animal model. Domestic pigs (n = 5) were subjected to contrast enhanced first pass MRI (MRI-FP) and the impulse response at different regions of the myocardium (n = 24/pig) were evaluated at rest (n = 120) and stress (n = 96). Reference MBF was then measured using positron emission tomography (PET). Since the impulse response may include artefacts, classification models based on SVM and RF were developed to discriminate noisy signal. In addition, regression models based on SVM, RF and linear regression (for comparison) were developed for estimating MBF from the impulse response at rest and stress. The classification and regression models were trained on data from 4 pigs (n = 168) and tested on 1 pig (n = 48). Models based on SVM and RF outperformed linear regression, with higher correlation (R SVM 2 = 0.81, R RF 2 = 0.74, R linear_regression 2 = 0.60; ρSVM = 0.76, ρRF = 0.76, ρlinear_regression = 0.71) and lower error (RMSESVM = 0.67 mL/g/min, RMSERF = 0.77 mL/g/min, RMSElinear_regression = 0.96 mL/g/min) for predicting MBF from MRI impulse response signal. Classifier based on SVM was optimal for detecting impulse response signals with artefacts (accuracy = 92%). Modified dual bolus MRI signal, combined with machine learning, has potential for accurately estimating MBF at rest and stress states, even from signals with dark rim artefacts. This could provide a protocol for reliable and easy estimation of MBF, although further research is needed to clinically validate the approach.

Quantification of porcine myocardial perfusion with modified dual bolus MRI - a prospective study with a PET reference.

BACKGROUND: The reliable quantification of myocardial blood flow (MBF) with MRI, necessitates the correction of errors in arterial input function (AIF) caused by the T1 saturation effect. The aim of this study was to compare MBF determined by a traditional dual bolus method against a modified dual bolus approach and to evaluate both methods against PET in a porcine model of myocardial ischemia. METHODS: Local myocardial ischemia was induced in five pigs, which were subsequently examined with contrast enhanced MRI (gadoteric acid) and PET (O-15 water). In the determination of MBF, the initial high concentration AIF was corrected using the ratio of low and high contrast AIF areas, normalized according to the corresponding heart rates. MBF was determined from the MRI, during stress and at rest, using the dual bolus and the modified dual bolus methods in 24 segments of the myocardium (total of 240 segments, five pigs in stress and rest). Due to image artifacts and technical problems 53% of the segments had to be rejected from further analyses. These two estimates were later compared against respective rest and stress PET-based MBF measurements. RESULTS: Values of MBF were determined for 112/240 regions. Correlations for MBF between the modified dual bolus method and PET was rs = 0.84, and between the traditional dual bolus method and PET rs = 0.79. The intraclass correlation was very good (ICC = 0.85) between the modified dual bolus method and PET, but poor between the traditional dual bolus method and PET (ICC = 0.07). CONCLUSIONS: The modified dual bolus method showed a better agreement with PET than the traditional dual bolus method. The modified dual bolus method was found to be more reliable than the traditional dual bolus method, especially when there was variation in the heart rate. However, the difference between the MBF values estimated with either of the two MRI-based dual-bolus methods and those estimated with the gold-standard PET method were statistically significant.

Association between bone mineral density and lumbar disc degeneration.

OBJECTIVES: Higher vertebral bone mineral density (BMD) has been found to be related with lumbar disc degeneration (LDD), while relationship between femoral neck BMD and LDD remains controversial. The aim of our research was to study the relationship between LDD and BMD of the lumbar spine and femoral neck. STUDY DESIGN: The study population consisted of 168 postmenopausal women (aged 63.3-75.0 years, mean 68.6 years) from the prospective OSTPRE and OSTPRE-FPS study cohorts. The severity of LDD was graded from T2-weighted MRI images using the five-grade Pfirrmann classification. Four vertebral levels (L1-L4) were studied (total 672 discs). The association between lumbar BMD and Z-score and the severity of LDD was studied separately for each vertebral level with AN(C)OVA analysis, using potential confounders as covariates. RESULTS: Higher lumbar BMD and Z-score were associated with more severe LDD at all studied levels (L1-L4): between L4-L5 disc and L4 BMD (p=0.044) and L4 Z-score (p=0.052), between L2-L3 disc and L3 BMD (p=0.001) and at all other levels (p<0.001). The mean degeneration grade of the studied discs was associated with the mean L1-L4 BMD and Z-score (p<0.001). Statistical significance of any result did not alter after controlling for confounding factors. There was no significant association between femoral neck BMD and LDD. CONCLUSIONS: Higher lumbar BMD/Z-score were associated with more severe LDD. There was no significant association between femoral neck BMD and disc degeneration. Femoral neck BMD may be a more reliable measurement for diagnosing osteoporosis in postmenopausal women with degenerative changes in the lumbar spine.

Assessment of splanchnic blood flow using magnetic resonance imaging.

BACKGROUND AND AIMS: The splanchnic circulation has an important function in the body under both physiological and pathophysiological conditions. Despite its importance, no reliable noninvasive procedures for estimating splanchnic circulation have been established. The aim of this study was to evaluate MRI as a tool for assessing intra-abdominal blood flows of the aorta, portal vein (VPO) and the major intestinal and hepatic vessels. METHODS: In nine healthy volunteers, the proximal aorta (AOP) and distal abdominal aorta (AOD), superior mesenteric artery (SAM), celiac trunk (CTR), hepatic arteries (common and proper hepatic arteries, AHC and AHP, respectively), and VPO were localized on contrast-enhanced magnetic resonance angiography images. Volumetric flow was measured using a two-dimensional cine echocardiogram-gated phase contrast technique. Measurements were taken before and 30 min after continuous intravenous infusion of somatostatin (250 microg/h) and were independently evaluated by two investigators. RESULTS: Blood flow measured by MRI in the VPO, SAM, AOP, AHP, and CTR significantly decreased after drug infusion. Flows in the AOD and AHC showed a tendency to decrease (P>0.05). Interrater agreement on flows in MRI was very good for large vessels (VPO, AOP, and AOD), with a concordance correlation coefficient of 0.94, as well as for smaller vessels such as the CTR, AHC, AHP, and SAM (concordance correlation coefficient =0.78). CONCLUSION: Somatostatin-induced blood flow changes in the splanchnic region were reliably detected by MRI. MRI may be useful for the noninvasive assessment of blood flow changes in the splanchnic region.

Progression and determinants of quantitative magnetic resonance imaging measures of lumbar disc degeneration: a five-year follow-up of adult male monozygotic twins.

STUDY DESIGN: A longitudinal study. OBJECTIVE: Our goal was to explore the role of digital magnetic resonance imaging (MRI) data, by extending our earlier 5-year follow-up study of progression of lumbar spine degeneration with quantitative measures of disc degeneration. SUMMARY OF BACKGROUND DATA: A longitudinal study is optimal for investigating disc degeneration but only a few studies (with small sample sizes) or short follow-up studies include repeated MRI data. METHODS: Subjects consisted of 134 male monozygotic twins (age 35-69 years). Quantitative MRI measures included changes in disc bulging and height. Inter-rater reliability coefficients were between 0.77 and 0.96. At baseline and follow-up, an extensive interview about exposures to suspected determinants was conducted. RESULTS: Reduction in disc height and increases in bulges (worsening) were seen in 2/3 of subjects. The mean reduction in disc height was 2.2% to 3.6%. A mean increase in bulging of 7% to 10% was found in the L1-L4 discs and 4% in L4-S1 discs. Although the mean changes were small, few reverse changes were observed. Familial aggregation, a proxy for genetic influences, explained 17% of changes in disc height, and 11% and 0% of changes in the sizes of anterior and posterior bulges in the regression models. Higher maximal occupational lifting (AR2 = 4.9%) and smoking (AR2 = 3.5%) during follow-up predicted more disc height reduction. Greater increases in bulging (AR2 = 7.4%-10.2%) were predicted by smaller bulges at baseline. CONCLUSION: The mean annual changes in disc heights (<1%) and bulges (<2%) were small, and included both decreases and increases, with only a few subjects showing more major changes in either direction. The role of genetics was largest except in posterior bulges, but lifting and smoking were also associated with disc height reduction but none of the other studied risk factors were associated with anterior or posterior disc bulging. Different degenerative findings have different determinants of progression.

Estimation of femoral head bone density using magnetic resonance imaging: comparison between men with and without hip osteoarthritis.

Bone changes are thought to be one important etiological element in the pathogenesis of hip osteoarthritis (OA). The magnetic resonance imaging (MRI)-derived T2* relaxation time has been shown to provide information about bone mineral status of the femoral neck. The aim of this study was to test the hypothesis that the MRI-derived T2* relaxation time of the proximal femur in hip OA differs from that seen in healthy subjects. Based on the American College of Rheumatology criteria regarding classification of the OA of the hip, 27 men (aged 47-64 yr) with unilateral or bilateral hip OA and 30 age-matched randomly selected healthy men were studied. Bone mineral density (BMD), bone mineral content (BMC), and bone width of the femoral neck were measured with dual-energy X-ray absorptiometry (DXA). Subsequently, T2* measurements were performed with a 1.5-T scanner (Siemens Magnetom 63SP; Erlangen, Germany). A single 10-mm-thick coronal slice was generated on the femur, with a repetition time of 60 ms and nine echo times (4-20 ms) to derive T2* values. T2* measurements were performed from the different region of interests (ROIs) from the femoral neck and head. T2* relaxation times showed significant negative correlations with BMC, BMD (r = -0.401 to -0.794; p < 0.05-0.001). T2* relaxation time values revealed no significant differences between the groups in the femoral neck and in the head of the femur, whereas it was 12% lower (p < 0.01) in OA subjects than in controls in acetabulum. There were no significant differences in the T2* relaxation time values between the radiographic OA subgroups. Our findings suggest that hip OA is not associated with an increase of BMD in the femoral neck or in the head of the femur.

Hippocampus and entorhinal cortex in mild cognitive impairment and early AD.

Magnetic resonance imaging (MRI) has been suggested as a useful tool in early diagnosis of Alzheimer's disease (AD). Based on MRI-derived volumes, we studied the hippocampus and entorhinal cortex (ERC) in 59 controls, 65 individuals with mild cognitive impairment (MCI) and 48 patients with AD. The controls and individuals with MCI were derived from population-based cohorts. Volumes of the hippocampus and ERC were significantly reduced in the following order: control > MCI > AD. Stepwise discriminant function analysis showed that the most efficient overall classification between controls and individuals with MCI subjects was achieved with ERC measurements (65.9%). However, the best overall classification between controls and AD patients (90.7%), and between individuals with MCI and AD patients (82.3%) was achieved with hippocampal volumes. Our results suggest that the ERC atrophy precedes hippocampal atrophy in AD. The ERC volume loss is dominant over the hippocampal volume loss in MCI, whereas more pronounced hippocampal volume loss appears in mild AD.

Adenovirus-mediated herpes simplex virus thymidine kinase gene therapy in BT4C rat glioma model.

Adenovirus (Adv)-mediated herpes simplex virus thymidine kinase (adv/tk) gene therapy combined with ganciclovir (GCV) medication is a promising approach for the treatment of malignant glioma. However, optimal administration and the effect of possible adjuvant treatments have not been fully examined. In the present study, we examined the efficacy of adv/tk/GCV gene therapy in a syngeneic BT4C rat malignant glioma model, either as a single administration or given as three injections during three consecutive days. The effect of combined adv-mediated macrophage colony-stimulating factor (MCSF) and adv/tk gene transfer was also studied. BT4C malignant glioma cells were injected into the right corpus callosum of BDIX rats (n=112). Before gene therapy, the presence of tumors was verified by MRI. The rats were divided into eight groups as follows: group I (n=20) received a single adv/tk gene transfer (total dose 4x10(8) pfu) and GCV treatment for 14 days; group II (n=5) received the same gene transfer without GCV; group III (n=28) received three adv/tk injections (total dose 4x10(8) pfu) on three consecutive days and GCV for 14 days; group IV (n=5) received three similar adv/tk injections without GCV medication; group V (n=13) received three adv/MCSF injections (total dose 2x10(8) pfu) on three consecutive days and GCV medication; group VI (n=12) received three adv/tk and adv/MCSF (total dose 6x10(8) pfu) injections on three consecutive days followed by GCV medication; and group VII (n=12) the same treatment without GCV. Group VIII (n=17) consisted of wild-type BT4C malignant glioma tumors without any treatment. Treatment effect and tissue responses were characterized by general histology, immunohistochemistry, MRI, and survival of the study groups. The best treatment effect and survival was found in rats treated with adv/tk gene transfer once a day for three consecutive days (P<.05). No improvement of the treatment effect was seen after the combined adv/tk and adv/MCSF gene transfer compared with the repeated adv/tk gene transfer. The results show that 20% of the rats can be cured (survival >6 months) after optimized adv/tk gene therapy. It is concluded that repeated intratumoral administration of adv/tk is a promising approach for the treatment of malignant glioma tumors in vivo.

Comparison of DXA and MRI methods for interpreting femoral neck bone mineral density.

The aim of the study was to improve the practical implementation of the dual X-ray absorptiometry (DXA) by converting the areal bone mineral density BMD (BMD(areal)) to volumetric BMD using magnetic resonance (MR) imaging (MRI) because a failure to control for the femoral neck size can lead to erroneous interpretation of BMD values. We also evaluated the feasibility of MR T2* relaxation time in assessing bone mineral status of the femoral neck. Twenty-eight randomly selected 47- to 64-yr-old healthy men were studied. The men had neither unilateral nor bilateral hip osteoarthritis according to radiographs. Bone width, mineral content (BMC), BMD(areal), and apparent volumetric BMD (BMD(vol)) of the right femoral neck were measured with DXA. The BMD(vol) was calculated by approximating the femoral neck to be cylindrical with a circular cross-section (Vol(dxa)). Volumetric measurements from MR (Vol(mri)) images of the femoral neck were also used to create a BMD measure that was corrected for the femoral neck volume (BMD(mri)). T2* measurements were performed with a 1.5-T scanner (Siemens Magnetom 63SP, Erlangen, Germany). A single 10-mm-thick coronal slice was generated on the femur with a repetition time of 60 ms, and nine echo times (4-20 ms) were used to derive T2* values. Vol(mri) correlated positively (r = 0.828, p < 0.001) with Vol(dxa). However, the Vol(mri) of the femoral neck was 18% lower than the Vol(dxa). Similarly, the BMD(mri) was related to the BMD(vol) (r = 0.737, p < 0.001). Because of the difference in the volumetric measures, the BMD(mri) of the femoral neck was 21% higher than the BMD(vol) (p < 0.001). T2* relaxation time showed a significant negative correlation with BMC, BMD(areal), BMD(vol), and BMD(mri) (r = -0.423 to -0.757, p < 0.05-0.001). In conclusion, these results are evidence that DXA-derived volume approximations by the cylinder with circular cross-section geometry may lead to lower DXA-derived BMD(vol) values, as compared to true MRI-derived volumetric bone mineral density. Thus, the BMD(vol) may not be an accurate method to calculate the true volumetric BMD in the femoral neck. Our results also suggest that the MRI-derived T2* method may be used to approximate the BMD in the proximal femur.