MyI-Net: Fully Automatic Detection and Quantification of Myocardial Infarction from Cardiovascular MRI Images

Myocardial infarction (MI) occurs when an artery supplying blood to the heart is abruptly occluded. The "gold standard" method for imaging MI is cardiovascular magnetic resonance imaging (MRI) with intravenously administered gadolinium-based contrast (with damaged areas apparent as late gadolinium enhancement [LGE]). However, no "gold standard" fully automated method for the quantification of MI exists. In this work, we propose an end-to-end fully automatic system (MyI-Net) for the detection and quantification of MI in MRI images. It has the potential to reduce uncertainty due to technical variability across labs and the inherent problems of data and labels. Our system consists of four processing stages designed to maintain the flow of information across scales. First, features from raw MRI images are generated using feature extractors built on ResNet and MoblieNet architectures. This is followed by atrous spatial pyramid pooling (ASPP) to produce spatial information at different scales to preserve more image context. High-level features from ASPP and initial low-level features are concatenated at the third stage and then passed to the fourth stage where spatial information is recovered via up-sampling to produce final image segmentation output into: (i) background, (ii) heart muscle, (iii) blood and (iv) LGE areas. Our experiments show that the model named MI-ResNet50-AC provides the best global accuracy (97.38%), mean accuracy (86.01%), weighted intersection over union (IoU) of 96.47%, and bfscore of 64.46% for the global segmentation. However, in detecting only LGE tissue, a smaller model, MI-ResNet18-AC, exhibited higher accuracy (74.41%) than MI-ResNet50-AC (64.29%). New models were compared with state-of-the-art models and manual quantification. Our models demonstrated favorable performance in global segmentation and LGE detection relative to the state-of-the-art, including a four-fold better performance in matching LGE pixels to contours produced by clinicians.

Cardiovascular structure and function assessed by MRI in healthy South Asians compared to White Europeans: a UK Biobank study

Background

There is limited data on ethnic specific comparisons for measures of cardiovascular structure and function in healthy cohorts. Echocardiographic data indicate South Asian's (SAs) have smaller mass and evidence of more concentric remodelling compared to White Europeans (WEs). Furthermore, there is no data published for strain or strain rates.

Purpose

To compare Cardiac Magnetic Resonance (CMR) derived measures of structure and function between age and sex matched healthy SAs and WEs from the UK biobank cohort.

Methods

South Asian and WE participants from the UK Biobank who underwent CMR imaging were included. Individuals with a history of cardiovascular disease, hypertension, obesity (BMI ≥30 kg/m2 in WEs and ≥27 kg/m2 in SAs) and diabetes were excluded. Ethnic groups were matched according to age and sex at recruitment. Cine images (bSSFP) were analysed blinded to participant details using commercially available software. Left ventricular (LV) mass, LV volumes, global longitudinal and circumferential systolic strain (GLS and GCS), together with peak early diastolic strain rates (PEDSR), were obtained. Data distributions were assessed and T-Test or Man Whitney U conducted as appropriate.

Results

Datasets from the UK biobank were screened (n=45000). After applying exclusion criteria, 111 pairs of matched SAs and WEs were available for analysis (n=69 male and n=42 female matched pairs). Mean age of the entire cohort was 58±8 years. Data has been corrected according to body surface area (BSA),(males: WE 2.0±0.1 vs SA 1.8±0.1 m2, p≤0.001; females: WE 1.7±0.2 vs SA 1.6±0.1 m2, p≤0.001). There was no difference in heart rate (males: WE 64.5±9.3 vs SA 65.8±9.6 bpm, p=0.113; females: WE 66.2±7.8 vs SA 69.5±10.3 bpm, p=0.125). Results by sex and ethnicity are displayed in table 1. In males there was no difference in ejection fraction (EF) or indexed LV end diastolic volume (LVEDV). However indexed mass and mass/volume ratio were significantly lower in SAs, and GLS but not GCS was significantly higher than in SAs. Longitudinal PEDSR were significantly higher in SAs than in WE. By contrast, SA females had significantly lower EF with no difference in indexed LVEDV compared to WE females. However, as seen with the males SA females had significantly lower indexed LV mass and mass/volume ratio compared to WE females. Finally, both GLS and GCS were significantly higher in SAs compared to WE females, whereas there was no difference in longitudinal PEDSR.

Conclusion

Substantial differences in cardiovascular structure and function exist between SA and WE ethnic groups, in both men and women. Contrary to previous echocardiographic studies, LV volumes were similar between ethnicities and SA have less, not increased, concentric remodelling than WE. These findings highlight the need for ethnicity and sex-specific healthy reference ranges derived from CMR.

Funding Acknowledgement

Type of funding sources: None.

The impact of lifestyle intervention on left atrial function in type 2 diabetes: results from the DIASTOLIC study

Aerobic exercise training and low energy diets have been shown to improve left ventricular remodelling and diastolic function in adults with type 2 diabetes (T2D), albeit with differential effects. The impact of these lifestyle interventions on left atrial (LA) function, however, has not previously been reported. The DIASTOLIC study was a prospective, randomised, open-label, blind endpoint trial, in which 90 people with obesity and T2D and no prevalent cardiovascular disease were randomised to a 12-week intervention of: (i) routine care, (ii) aerobic exercise training, or (iii) low energy (≈ 810 kcal/day) meal replacement plan (MRP). Cardiac magnetic resonance (CMR) imaging was performed pre- and post-intervention. Image analysis included LA volumes (LAV), emptying fraction (LAEF), and LA strain (LAS) corresponding to LA reservoir (LAS-r), conduit (LAS-cd), and booster pump (LAS-bp) function. 73 participants with T2D (mean age 50 ± 6 years, 62% male, body mass index (BMI) 36.1 ± 5.3 kg/m2) completed the trial and had analysable LA images. There was no significant change in CMR measured LA volumetric function (LAV/LAEF) in any group. The routine care group showed no significant change in BMI or LAS. In the MRP group, there were significant reductions in BMI (4.5 kg/m2) and a significant increase in LAS-r and LAS-bp (29.9 ± 7.0 to 32.3 ± 7.0%, p = 0.036 and 14.6 ± 5.3 to 17.2 ± 3.7%, p = 0.034). The exercise group showed a small reduction in BMI (0.49 kg/m2), with no significant change in LAS. Compared to routine care, weight loss via a 12-week MRP, led to improvements in LA filling and contractile function in adults with T2D and obesity. However, these within-group changes were not statistically significant on between-group comparison. ClinicalTrials.gov Identifier: NCT02590822.

Fibro-inflammatory recovery and type 2 diabetes remission following a low calorie diet but not exercise training: A secondary analysis of the DIASTOLIC randomised controlled trial

AIMS

To investigate the relationship between fibro-inflammatory biomarkers and cardiovascular structure/function in people with Type 2 Diabetes (T2D) compared to healthy controls and the effect of two lifestyle interventions in T2D.

METHODS

Data were derived from the DIASTOLIC randomised controlled trial (RCT) and includes a comparison between those with T2D and the matched healthy volunteers recruited at baseline. Adults with T2D without cardiovascular disease (CVD) were randomized to a 12-week intervention either: (1) exercise training, (2) a low-energy (∼810 kcal/day) meal-replacement plan (MRP) or (3) standard care. Principal Component and Fisher's linear discriminant analysis were used to investigate the relationships between MRI acquired cardiovascular outcomes and fibro-inflammatory biomarkers in cases versus controls and pre- and post-intervention in T2D.

RESULTS

At baseline, 83 people with T2D (mean age 50.5 ± 6.4; 58% male) and 36 healthy controls (mean age 48.6 ± 6.2; 53% male) were compared and 76 people with T2D completed the RCT for pre- post-analysis. Compared to healthy controls, subjects with T2D had adverse cardiovascular remodelling and a fibro-inflammatory profile (20 differentially expressed biomarkers). The 3D data visualisations showed almost complete separation between healthy controls and those with T2D, and a marked shift towards healthy controls following the MRP (15 biomarkers significantly changed) but not exercise training.

CONCLUSIONS

Fibro-inflammatory pathways and cardiovascular structure/function are adversely altered before the onset of symptomatic CVD in middle-aged adults with T2D. The MRP improved the fibro-inflammatory profile of people with T2D towards a more healthy status. Long-term studies are required to assess whether these changes lead to continued reverse cardiac remodelling and prevent CVD.

Test-retest reproducibility of echocardiography and CMR in the same patients for assessing left atrial function

Funding Acknowledgements

Type of funding sources: Other. Main funding source(s): King Saud bin Abdulaziz University for Health Sciences, Riyadh, Saudi Arabia

Background

 Cardiovascular magnetic resonance (CMR) has been shown to give higher left atrial (LA) volumes than trans-thoracic echocardiography (TTE) but the agreement and test-retest reproducibility of LA functional measure have not been directly compared in the same patient cohort.

Methods

 People with type 2 diabetes mellitus (T2D) without cardiovascular disease underwent TTE and CMR on the same day on two separate occasions with a mean period of 11± 4 days. TTE images were analysed using TomTec-ARENA (v2.4, 2D-CPA), where Medis Suite (v3.1, medical imaging system) was used for CMR images. The analysis included LA strain, corresponding to LA reservoir, conduit, and booster pump (contraction function), LA volumes, and LA emptying fraction (LAEF), which was calculated using area length method. Strain and volume were assessed on 4- and 2-chamber long-axis and average values were calculated for both modalities.

Results

 10 participants with T2D (mean age 65.6 ± 7.3 years, 50% male) were studied. CMR LA volumes were significantly higher and LAEF significantly lower compared to TTE (P < 0.01 for all), whilst reservoir strain on CMR was lower compared to TTE (29.2 ± 6.5 vs 33.8 ± 3.7, p = 0.04).  The inter-modality agreement was moderate for LA volumes and contraction strain (ICC ≥0.55) and poor for LAEF, strain at reservoir and conduit. Overall, the test-retest reproducibility of CMR was higher than for TTE, but the overall limits of agreement were not too dissimilar on Bland-Altman analysis (Table & Figure). TTE reproducibility was good for LA volumes and strain at conduit (CoV 18-24%, ICC ∼0.80) and moderate for LAEF, strain at reservoir and contraction. CMR showed excellent test-retest reproducibility for LA volumes, EF and strain at reservoir (CoV 7-20%, ICC ≥0.83).

Conclusion

 The test-retest reproducibility of CMR is slightly better than TTE for LA assessment using volumes and strain. LA volumes have superior reproducibility compared to functional LA parameters. Abstract Table: Test-retest reproducibility  Abstract Figure: Bland-Altman plot

Lifestyle intervention improves Left atrial strain in type 2 diabetes: results from the DIASTOLIC randomised trial

Introduction

Left atrial (LA) strain has been shown to be an independent predictor of cardiovascular events in many conditions and superior to conventional measures of LA function (LA volumes and emptying fraction). The effects of low-energy diet or exercise on cardiovascular function in younger adults with type 2 diabetes (DIASTOLIC) study has previously demonstrated an improvement in left ventricle peak early diastolic strain rate in response to aerobic exercise in patients with type-2 diabetes (T2D). However, the effect of lifestyle intervention on LA function, assessed by cardiac magnetic resonance (CMR) imaging, is unknown.

Methods

The DIASTOLIC study was a prospective, randomised, open-label, blind endpoint trial, that randomised 90 obese participants with T2D (aged 18–65 years) to a 12-week intervention of: (i) routine care, (ii) aerobic exercise training, or (iii) low energy (≈810kcal/day) meal replacement plan (MRP). CMR was performed at baseline and week-12. Images were analysed using Medis v3.1. LA strain (LAS) was assessed using Feature Tracking (QStrain v2.0), corresponding to LA reservoir (LAS-r), conduit (LAS-cd), and booster pump (LAS-bp) using 4- and 2-chamber standard steady-state free precession cine images, and average values calculated. LA emptying fraction (LAEF) was calculated using biplane area-length method (QMass v8.1) for total, passive and active EF.

Results

73 participants with T2D completed the trial and had analysable LA images (28 routine care, 22 exercise and 23 MRP). There was no significant change in CMR measured standard LA volumetric function (LAV/LAEF) and echocardiogram measured LV filling pressures (E/e') in any groups. The routine care arm showed no significant change in body mass index (BMI) or LAS (see Table). In the MRP group, there were significant reductions in BMI (4.8 kg/m2), mean systolic blood pressure (SBP) (13mmHg), and a significant increase in LAS-r and LAS-bp (29.9±7.0 to 32.3±7.0,p=0.036 and 14.6±5.3 to 17.2±3.7,p=0.034). The exercise arm showed a small reduction in BMI (0.8kg/m2), with no significant change in BP or LAS.

Conclusion

There was significant improvement in LAS post-lifestyle intervention in young adults with T2D, despite no change in volumetric measurements. MRP led to weight loss and improved SBP, with associated improved LA filling and contraction.

Funding Acknowledgement

Type of funding sources: Public grant(s) – National budget only. Main funding source(s): National Institute for Health Research (NIHR) a career development fellowship.

Interrelationship between micronutrients and cardiovascular structure and function in type 2 diabetes

Micronutrients are important for normal cardiovascular function. They may play a role in the increased risk of cardiovascular disease observed in people with type 2 diabetes (T2D) and T2D-related heart failure. The aims of this study were to (1) examine micronutrient status in people with T2D v. healthy controls; (2) assess any changes following a nutritionally complete meal replacement plan (MRP) compared with routine care; (3) determine if any changes were associated with changes in cardiovascular structure/function. This was a secondary analysis of data from a prospective, randomised, open-label, blinded end-point trial of people with T2D, with a nested case-control [NCT02590822]. Anthropometrics, cardiac resonance imaging and fasting blood samples (to quantify vitamins B1, B6, B12, D and C; and iron and ferritin) were collected at baseline and 12 weeks following the MRP or routine care. Comparative data in healthy controls were collected at baseline. A total of eighty-three people with T2D and thirty-six healthy controls were compared at baseline; all had micronutrient status within reference ranges. Vitamin B1 was higher (148⋅9 v. 131⋅7; P 0⋅01) and B6 lower (37⋅3 v. 52⋅9; P 0⋅01) in T2D v. controls. All thirty participants randomised to routine care and twenty-four to the MRP completed the study. There was an increase in vitamins B1, B6, D and C following the MRP, which were not associated with changes in cardiovascular structure/function. In conclusion, changes in micronutrient status following the MRP were not independently associated with improvements in cardiovascular structure/function in people with T2D.

Differences in native T1 and native T2 mapping between patients on hemodialysis and control subjects

PURPOSE

Myocardial native T1 is a potential measure of myocardial fibrosis, but concerns remain over the potential influence of myocardial edema to increased native T1 signal in subjects prone to fluid overload. This study describes differences in native T2 (typically raised in states of myocardial edema) and native T1 times in patients on hemodialysis by comparing native T1 and native T2 times between subjects on hemodialysis to an asymptomatic control group. Reproducibility of these sequences was tested.

METHODS

Subjects were recruited prospectively and underwent 3 T-cardiac MRI with acquisition of native T1 and native T2 maps. Between group differences in native T1 and T2 maps were assessed using one-way ANOVAs. 30 subjects underwent test-retest scans within a week of their original scan to define sequence reproducibility.

RESULTS

261 subjects completed the study (hemodialysis n = 124, control n = 137). Native T1 times were significantly increased in subjects on hemodialysis compared to control subjects (1259 ms ± 51 vs 1212 ms ± 37, p < 0.01). There was no difference in native T2 times between subjects on hemodialysis and control subjects (39.5 ms ± 2.5 vs 39.5 ms ± 2.3, p = 0.9). These differences were unchanged after adjustment for relevant baseline differences (age, sex and hemoglobin). Inter-study reproducibility for native T1 and T2 mapping was excellent (coefficient of variability 0.9 % and 2.6 % respectively).

CONCLUSIONS

The increased native T1 signal demonstrated in subjects on hemodialysis occurs independently of differences in native T2 and the two parameters are not orthogonal. Elevated native T1 in patients on hemodialysis may be driven by water related to myocardial fibrosis rather than edema from volume overload.

Pathogens associated with pleuritic pig lungs at an abattoir in Queensland Australia

OBJECTIVE

Pleurisy in pigs has economic impacts in the production stage and at slaughter. This study sought to establish if some micro-organisms can be found in high numbers in lungs with pleurisy by assessing batches of pigs at an abattoir in Queensland Australia.

DESIGN

Samples of lung (including trachea/bronchus and lymph nodes) from a maximum of 5 pleurisy affected pigs were collected from 46 batches of pigs representing 46 Queensland farms.

PROCEDURE

Pleurisy-affected lung areas were cultured by traditional bacteriological methods and bacteria quantified by plate scores. Additionally, tracheal or bronchial swabs and apical lobe fluid were tested for Mycoplasma hyopneumoniae DNA and the superior tracheobronchial lymph nodes were tested for porcine circovirus type 2 DNA by polymerase chain reaction (PCR). All apparently significant bacteria were identified via PCR or sequencing. Typing was undertaken on some of the bacterial isolates.

RESULTS

The most prevalent pathogens were M. hyopneumoniae, Streptococcus suis and Porcine Circovirus type 2, being found in 34, 38 and 31 batches, respectively. Other bacteria found were Actinobacillus species (29 batches), Pasteurella multocida (24 batches), Mycoplasma flocculare (9 batches), Actinobacillus pleuropneumoniae (7 batches), Mycoplasma hyorhinis (4 batches), Bisgaard Taxon 10 (1 batch), Glaesserella parasuis (1 batch), Streptococcus minor (1 batch) and Streptococcus porcinus (1 batch). Most batches had more than one bacterial species.

CONCLUSION

The high percentage of batches infected with S. suis (83%), M. hyopneumoniae (74%) and PCV2 (70%) and clustering by a batch of these pathogens, as well as the presence of many secondary pathogens, suggests synergy between these organisms may have resulted in pleurisy.

The reproducibility of cardiac magnetic resonance imaging measures of aortic stiffness and their relationship to cardiac structure in prevalent haemodialysis patients

BACKGROUND

Aortic stiffness is one of the earliest signs of cardiovascular disease (CVD) in patients with chronic kidney disease and an independent predictor of mortality. It is thought to drive left ventricular (LV) remodelling, an established biomarker for mortality. The relationship between direct and indirect measures of aortic stiffness and LV remodelling is not defined in dialysis patients, nor are the reproducibility of methods used to assess aortic stiffness using cardiac magnetic resonance (CMR) imaging.

METHODS

Using 3T CMR, we report the results of (i) the interstudy, interobserver and intra-observer reproducibility of ascending aortic distensibility (AAD), descending aortic distensibility (DAD) and aortic pulse wave velocity (aPWV) in 10 haemodialysis (HD) patients and (ii) the relationship between AAD, DAD and aPWV and LV mass index (LVMi) and LV remodelling in 70 HD patients.

RESULTS

Inter- and intra-observer variability of AAD, DAD and aPWV were excellent [intraclass correlation (ICC) > 0.9 for all]. Interstudy reproducibility of AAD was excellent {ICC 0.94 [95% confidence interval (CI) 0.78-0.99]}, but poor for DAD and aPWV [ICC 0.51 (-0.13-0.85) and 0.51 (-0.31-0.89)]. AAD, DAD and aPWV associated with LVMi on univariate analysis (β = -0.244, P = 0.04; β =-0.315, P < 0.001 and β = 0.242, P = 0.04, respectively). Only systolic blood pressure, serum phosphate and a history of CVD remained independent determinants of LVMi on multivariable linear regression.

CONCLUSIONS

AAD is the most reproducible CMR-derived measure of aortic stiffness in HD patients. CMR-derived measures of aortic stiffness were not independent determinants of LVMi in HD patients. Whether one should target blood pressure over aortic stiffness to mitigate cardiovascular risk still needs determination.