Exploring the role of artificial intelligence in the study of fetal heart

The International Journal of Cardiovascular Imaging -     

Application of automated sonography on 4-dimensional volumes of fetuses with transposition of the great arteries

OBJECTIVE: The purpose of this study was to assess, in second-trimester fetuses with transposition of the great arteries (TGA), the performance of software (sonographically based volume computer-aided analysis) that automatically retrieves diagnostic cardiac planes from a 4-dimensional volume of the fetal chest obtained with spatiotemporal image correlation. METHODS: We retrospectively evaluated the 4-dimensional spatiotemporal image correlation volumes of 12 fetuses with TGA (complete TGA, 10 cases; correct TGA, 2 cases). The data were analyzed to determine whether the target diagnostic planes, that is, cardiac plane 1 (left ventricle outflow tract) and cardiac plane 2 (right ventricle outflow tract), were correctly identified in at least 1 of the 7 automatically generated tomographic sonographic image displays and whether they allowed diagnosis of TGA. RESULTS: In 9 of 10 fetuses with complete TGA, target diagnostic cardiac plane 1 showed a branching arterial vessel (pulmonary artery) arising from the left ventricle, whereas in 7 of 10 fetuses, the aorta arising from the right ventricle was shown. In both cases with correct TGA, the pulmonary artery starting from the morphologic left ventricle was shown, whereas in 1 of 2, the connection of the aorta with the morphologic right ventricle was found. In all of the fetuses with TGA, a ventricular arterial connection anomaly was shown in either cardiac plane 1 or 2. CONCLUSIONS: This automatic approach shows good retrieval of diagnostic cardiac planes in fetuses with TGA, which may improve diagnostic efficacy for this disease.     

Experimental First-Pass Method for Testing and Comparing Sorbent Polymers Used in the Clearance of Iodine Contrast Materials

Background: Sorbents have been shown to adsorb iodinated radiocontrast media. Objective: In this study we describe a simple method to compare various sorbents in terms of capacity to adsorb radiocontrast media. Methods: Iodixanol solution was injected into columns filled with three types of sorbent at filtration velocities of increasing magnitude. Two variables of interest - contrast removal rate and matched iodine retention (MIR) - were calculated to measure the adsorption efficiency and the mass of contrast iodine adsorbed versus sorbent used, respectively. Results: The highest contrast removal and MIR for Porapak Q, CST 401 and Amberlite XAD4 were 41, 38 and 16% (p = 0.22 and 0.0005 for comparisons between Porapak Q-CST 401 and CST 401-Amberlite XAD4) and 0.060, 0.055 and 0.024, respectively (p = 0.18 and 0.0008). Extrapolation to a clinical scenario may suggest that removal of 8 ml iodixanol could be achieved by masses of sorbents of 43, 47 and 107 g, respectively. Conclusion: In this study we set a benchmark for comparing the radiocontrast-adsorbing efficiency of polymer sorbents during first-pass experiments, using a readily available methodology.     

Increased expression of renal cyclooxygenase-2 and neuronal nitric oxide synthase in hypertensive Cx40-deficient mice

Cx40-deficient mice (Cx40-/-) are hypertensive due to increased renin secretion. We evaluated the renal expression of neuronal nitric oxide synthase (nNOS) and cyclooxygenases COX-1 and COX-2, three macula densa enzymes. The levels of nNOS were increased in kidneys of Cx40-/- mice, as well as in those of wild-type (WT) mice subjected to the two-kidney one-clip model of hypertension. In contrast, the levels of COX-2 expression were only increased in the hypoperfused kidney of Cx40-/- mice. Treatment with indomethacin lowered blood pressure and renin mRNA in Cx40-/- mice without affecting renin levels, indicating that changes in COX-2 do not cause the altered secretion of renin. Suppression of NOS activity by N(G)-nitro-L-arginine methyl ester (L-NAME) decreased renin levels in Cx40-/- animals, indicating that NO regulates renin expression in the absence of Cx40. Treatment with candesartan normalized blood pressure in Cx40-/- mice, and decreased the levels of both COX-2 and nNOS. After a treatment combining candesartan and L-NAME, the blood pressure of Cx40-/- mice was higher than that of WT mice, showing that NO may counterbalance the vasoconstrictor effects of angiotensin II in Cx40-/- mice. These data document that renal COX-2 and nNOS are differentially regulated due to the elevation of renin-dependent blood pressure in mice lacking Cx40.     

Expression and modulation of steroidogenic acute regulatory protein messenger ribonucleic acid in rat cardiocytes and after myocardial infarction

We examined whether the mRNA for steroidogenic acute regulatory (StAR) protein, a crucial factor in the rate-limiting step of aldosterone biosynthesis, is expressed and regulated in rat heart. We performed quantitative RT-PCR for StAR mRNA in an in vitro and an in vivo model: purified rat neonatal cardiomyocytes in primary culture and myocardial infarction (MI) in the rat. StAR mRNA was expressed in cultured cardiomyocytes, and angiotensin II (10 nM) increased it in a time-dependent fashion (132 +/- 2.7% of controls after 24 h; n = 3; P < 0.05). Concomitantly, angiotensin II stimulated aldosterone production in the culture medium from 32.6 +/- 6.1 to 54 +/- 12.7 fmol/mg protein after 24 h (n = 8; P < 0.05). StAR mRNA levels in cardiomyocytes were dramatically reduced after 24-h treatment with dexamethasone in a concentration-dependent manner (50% inhibitory concentration, 10 nM); maximal inhibition (to 15 +/- 6% of control; P < 0.001; n = 6) was achieved with 100 nM dexamethasone. This inhibition was prevented by RU486. In the rat MI model, StAR mRNA was also present in control heart tissue and was increased 2.4-fold in the noninfarcted area of the left ventricle after MI (n = 6; P < 0.01). This effect was completely prevented by treatment with losartan (8 mg/kg per d) and spironolactone (80 mg/kg per d), which reduced StAR mRNA levels to values not different from those in non-MI controls. Thus, the mRNA for an indispensable factor in aldosterone biosynthesis, the StAR protein, is expressed in the rat heart and is up-regulated after MI. These results support the view of a local synthesis of aldosterone in the heart and of intracrine/paracrine deleterious effects of the mineralocorticoid in heart failure.     

Corticotropin-releasing activity of the renin-angiotensin system peptides in rat and in man

In this study we investigated in the rat the binding and corticotropin-releasing factor (CRF) activity of various constituents of the renin-angiotensin system and the possible angiotensin II receptor changes following procedures known to alter plasma renin activity. We investigated also the CRF activity of angiotensin II in vitro and in vivo in humans. The CRF activity of peptides was studied by their ability to stimulate ACTH release from pituitary cells. Deleting amino acids from the N-terminus of angiotensin II resulted in decreased CRF activity; while the ED50 for angiotensin II was 2 nM, it increased to about 10 nM for the (2-8)-heptapeptide. Angiotensin I had a weak CRF activity, whereas the substrate angiotensinogen had no stimulatory effect even at a concentration of 100 nM. There was a strong correlation between the activation and binding properties of all peptides tested. Dietary salt load or depletion as well as dexamethasone treatment did not affect the number nor the affinity of pituitary angiotensin II receptors. Angiotensin II had a CRF activity on human pituitary cells in vitro. However, peripherally injected agiotensin II at a pressive dose of 7 ng/kg/min did not produce any ACTH release in normal male volunteers. These data suggest that angiotensin II may play a modulatory role in the physiological regulation of ACTH secretion, but this role might be attributed to the endogenous brain angiotensin II as it is not closely dependent on the angiotensin II plasma levels.     

Axonal Neuropathies due to Mutations in Small Heat Shock Proteins: Clinical,Genetic, and Functional Insights into Novel Mutations

In this study, we describe the phenotypic spectrum of distal hereditary motor neuropathy caused by mutations in the small heat shock proteins HSPB1 and HSPB8 and investigate the functional consequences of newly discovered variants. Among 510 unrelated patients with distal motor neuropathy, we identified mutations in HSPB1 (28 index patients/510; 5.5%) and HSPB8 (four index patients/510; 0.8%) genes. Patients have slowly progressive distal (100%) and proximal (13%) weakness in lower limbs (100%), mild lower limbs sensory involvement (31%), foot deformities (73%), progressive distal upper limb weakness (29%), mildly raised serum creatine kinase levels (100%), and central nervous system involvement (9%). We identified 12 HSPB1 and four HSPB8 mutations, including five and three not previously reported. Transmission was either dominant (78%), recessive (3%), or de novo (19%). Three missense mutations in HSPB1 (Pro7Ser, Gly53Asp, and Gln128Arg) cause hyperphosphorylation of neurofilaments, whereas the C‐terminal mutant Ser187Leu triggers protein aggregation. Two frameshift mutations (Leu58fs and Ala61fs) create a premature stop codon leading to proteasomal degradation. Two mutations in HSPB8 (Lys141Met/Asn) exhibited increased binding to Bag3. We demonstrate that HSPB1 and HSPB8 mutations are a major cause of inherited motor axonal neuropathy. Mutations lead to diverse functional outcomes further demonstrating the pleotropic character of small heat shock proteins.     

Role of tomographic ultrasound imaging with spatiotemporal image correlation for identifying fetal ventricular septal defects

OBJECTIVE: The purpose of this study was to test the possibility of diagnosing ventricular septal defects (VSDs) by using tomographic ultrasound imaging (TUI) on 4-dimensional volumes of the fetal chest obtained with spatiotemporal image correlation and the color Doppler function. METHODS: We retrospectively evaluated the 4-dimensional spatiotemporal image correlation volumes of 8 fetuses with VSDs (single in 7 and double in 1) between 20 and 33 weeks' gestation. The data were analyzed to determine whether VSDs were correctly identified in at least 1 of 7 automatically generated TUI displays. RESULTS: All of the included VSDs were correctly identified with the use of an interslice distance in the TUI function ranging from 0.8 to 2 mm without the need to further manipulate the volume. CONCLUSIONS: This automatic approach shows good retrieval of diagnostic cardiac planes in fetuses with VSDs, which may improve the diagnostic accuracy for this disease.