Deep learning models have been effective for various fetal ultrasound segmentation tasks. However, generalization to new unseen data has raised questions about their effectiveness for clinical adoption. Normally, a transition to new unseen data requires time-consuming and costly quality assurance processes to validate the segmentation performance post-transition. Segmentation quality assessment efforts have focused on natural images, where the problem has been typically formulated as a dice score regression task. In this paper, we propose a simplified Fetal Ultrasound Segmentation Quality Assessment (FUSQA) model to tackle the segmentation quality assessment when no masks exist to compare with. We formulate the segmentation quality assessment process as an automated classification task to distinguish between good and poor-quality segmentation masks for more accurate gestational age estimation. We validate the performance of our proposed approach on two datasets we collect from two hospitals using different ultrasound machines. We compare different architectures, with our best-performing architecture achieving over 90% classification accuracy on distinguishing between good and poor-quality segmentation masks from an unseen dataset. Additionally, there was only a 1.45-day difference between the gestational age reported by doctors and estimated based on CRL measurements using well-segmented masks. On the other hand, this difference increased and reached up to 7.73 days when we calculated CRL from the poorly segmented masks. As a result, AI-based approaches can potentially aid fetal ultrasound segmentation quality assessment and might detect poor segmentation in real-time screening in the future.

To investigate metabolic changes of mild cognitive impairment in Parkinson’s disease (PD-MCI) using proton magnetic resonance spectroscopic imaging (1H-MRSI). Sixteen healthy controls (HC), 26 cognitively normal Parkinson’s disease (PD-CN) patients, and 34 PD-MCI patients were scanned in this prospective study. Neuropsychological tests were performed, and three-dimensional 1H-MRSI was obtained at 3 T. Metabolic parameters and neuropsychological test scores were compared between PD-MCI, PD-CN, and HC. The correlations between neuropsychological test scores and metabolic intensities were also assessed. Supervised machine learning algorithms were applied to classify HC, PD-CN, and PD-MCI groups based on metabolite levels. PD-MCI had a lower corrected total N-acetylaspartate over total creatine ratio (tNAA/tCr) in the right precentral gyrus, corresponding to the sensorimotor network (p = 0.01), and a lower tNAA over myoinositol ratio (tNAA/mI) at a part of the default mode network, corresponding to the retrosplenial cortex (p = 0.04) than PD-CN. The HC and PD-MCI patients were classified with an accuracy of 86.4

To assess fetal health during pregnancy, doctors use the gestational age (GA) calculation based on the Crown Rump Length (CRL) measurement in order to check for fetal size and growth trajectory. However, GA estimation based on CRL, requires proper positioning of calipers on the fetal crown and rump view, which is not always an easy plane to find, especially for an inexperienced sonographer. Finding a slightly oblique view from the true CRL view could lead to a different CRL value and therefore incorrect estimation of GA. This study presents an AI-based method for a quality assessment of the CRL view by verifying 7 clinical scoring criteria that are used to verify the correctness of the acquired plane. We show how our proposed solution achieves high accuracy on the majority of the scoring criteria when compared to an expert. We also show that if such scoring system is used, it helps identify poorly acquired images accurately and hence may help sonographers acquire better images which could potentially lead to a better assessment of conditions such as Intrauterine Growth Restriction (IUGR).

Fetal gestational age (GA) is vital clinical information that is estimated during pregnancy in order to assess fetal growth. This is usually performed by measuring the crown-rump-length (CRL) on an ultrasound image in the Dating scan which is then correlated with fetal age and growth trajectory. A major issue when performing the CRL measurement is ensuring that the image is acquired at the correct view, otherwise it could be misleading. Although clinical guidelines specify the criteria for the correct CRL view, sonographers may not regularly adhere to such rules. In this paper, we propose a new deep learning-based solution that is able to verify the adherence of a CRL image to clinical guidelines in order to assess image quality and facilitate accurate estimation of GA. We first segment out important fetal structures then use the localized structures to perform a clinically-guided mapping that verifies the adherence of criteria. The segmentation method combines the benefits of Convolutional Neural Network (CNN) and the Vision Transformer (ViT) to segment fetal structures in ultrasound images and localize important fetal landmarks. For segmentation purposes, we compare our proposed work with UNet and show that our CNN/ViT-based method outperforms an optimized version of UNet. Furthermore, we compare the output of the mapping with classification CNNs when assessing the clinical criteria and the overall acceptability of CRL images. We show that the proposed mapping is not only explainable but also more accurate than the best performing classification CNNs.

Proton magnetic resonance spectroscopic imaging ( 1 H‐MRSI) provides a noninvasive, spatially resolved evaluation of brain metabolism. However, there are some limitations of 1 H‐MRSI preventing its wider use in the clinics, including the spectral quality issues, partial volume effect and chemical shift artifact. Additionally, it is necessary to create metabolite maps for analyzing spectral data along with other MRI modalities. In this study, a MATLAB‐based open‐source data analysis software for three‐dimensional 1 H‐MRSI, called Oryx‐MRSI, which includes modules for visualization of raw 1 H‐MRSI data and LCModel outputs, chemical shift correction, tissue fraction calculation, metabolite map production, and registration onto standard MNI152 brain atlas while providing automatic spectral quality control, is presented. Oryx‐MRSI implements region of interest analysis at brain parcellations defined on MNI152 brain atlas. All generated metabolite maps are stored in NIfTI format. Oryx‐MRSI is publicly available at https://github.com/Computational-Imaging-LAB/Oryx-MRSI along with six example datasets.

Mild cognitive impairment of Parkinson's disease (PD) may be an early manifestation that may progressively worsen to dementia. Cognitive decline has been associated with changes in the brain perfusion pattern. This study aimed to evaluate cerebral blood flow (CBF) deficits specific to different stages of cognitive decline. Seventeen patients with cognitively normal PD (PD-CN), 18 patients with PD with mild cognitive impairment (PD-MCI), and 16 patients with PD with dementia (PDD) were included in this study. The participants were scanned using a 3 T Philips MRI scanner. Arterial spin labelling magnetic resonance (ASL-MR) images were acquired, followed by calculation of the CBF maps, and registration onto the MNI152 brain atlas. A whole-brain voxel-based CBF comparison was performed among the patient groups using age as a covariate. The mean age of patients with PDD was significantly higher than that of patients with PD-MCI (P = 0.015) and PD-CN (P = 0.001). The CBF values of the three groups were significantly different in the left cuneus of the visual network (VN), left inferior frontal gyrus of the frontoparietal network (FPN), and left dorsomedial nucleus of the thalamus. PDD had lower perfusion values than PD-MCI group in the same regions detected in the main group analysis. Additionally, comparison of PDD with PD-CN and non-demented groups revealed that the perfusion reduction extended into the bilateral cuneus of the VN, bilateral thalami, and left inferior frontal gyrus of the FPN. PDD could be separated from PD-MCI and PD-CN stages with CBF deficits in non-dopaminergically mediated posterior and dopaminergically mediated frontal networks.

Background There is a growing interest in noninvasively defining molecular subsets of hemispheric diffuse gliomas based on the isocitrate dehydrogenase ( IDH ) and telomerase reverse transcriptase gene promoter ( TERTp ) mutation status, which correspond to distinct tumor entities, and differ in demographics, natural history, treatment response, recurrence, and survival patterns. Purpose To investigate whether metabolite levels detected with short echo time (TE) proton MR spectroscopy ( 1 H‐MRS) at 3T can be used for noninvasive molecular classification of IDH and TERTp mutation‐based subsets of gliomas. Study Type Retrospective. Subjects In all, 112 hemispheric diffuse gliomas (70 males/42 females, mean age: 42.1 ± 13.9 years). Field Strength/Sequence Short‐TE 1 H‐MRS (repetition time (TR) = 2000 msec, TE = 30 msec, number of signal averages = 192) and routine clinical brain tumor MR protocols were acquired at 3T. Assessment 1 H‐MRS data were quantified using LCModel software. TERTp and IDH1 or IDH2 ( IDH1/2 ) mutations in the tissue were determined by either minisequencing or Sanger sequencing. Statistical Tests Metabolic differences between IDH mutant and IDH wildtype gliomas were assessed by a Mann–Whitney U ‐test. A Kruskal–Wallis test followed by a Tukey–Kramer test was used to analyze metabolic differences between IDH and TERTp mutational molecular subsets of gliomas. A Spearman rank correlation coefficient was used to assess the correlations of metabolite intensities with the Ki‐67 index. Furthermore, machine learning was employed to classify the IDH and TERTp mutational status of gliomas, and the accuracy, sensitivity, and specificity values were estimated. Results Short‐TE 1 H‐MRS classified the presence of an IDH mutation with 88.39% accuracy, 76.92% sensitivity, and 94.52% specificity, and a TERTp mutation within primary IDH wildtype gliomas with 92.59% accuracy, 83.33% sensitivity, and 95.24% specificity. Data Conclusion Short‐TE 1 H‐MRS could be used to identify molecular subsets of hemispheric diffuse gliomas corresponding to IDH and TERTp mutations. Level of Evidence: 3 Technical Efficacy Stage: 2 J. Magn. Reson. Imaging 2020;51:1799–1809.

Parkinson’s disease (PD) with mild cognitive impairment (PD-MCI) is currently diagnosed based on an arbitrarily predefined standard deviation of neuropsychological test scores, and more objective biomarkers for PD-MCI diagnosis are needed. The purpose of this study was to define possible brain perfusion-based biomarkers of not only mild cognitive impairment, but also risky gene carriers in PD using arterial spin labeling magnetic resonance imaging (ASL-MRI). Fifteen healthy controls (HC), 26 cognitively normal PD (PD-CN), and 27 PD-MCI subjects participated in this study. ASL-MRI data were acquired by signal targeting with alternating radio-frequency labeling with Look–Locker sequence at 3 T. Single nucleotide polymorphism genotyping for rs9468 [microtubule-associated protein tau (MAPT) H1/H1 versus H1/H2 haplotype] was performed using a Stratagene Mx3005p real-time polymerase chain-reaction system (Agilent Technologies, USA). There were 15 subjects with MAPT H1/H1 and 11 subjects with MAPT H1/H2 within PD-MCI, and 33 subjects with MAPT H1/H1 and 19 subjects with MAPT H1/H2 within all PD. Voxel-wise differences of cerebral blood flow (CBF) values between HC, PD-CN and PD-MCI were assessed by one-way analysis of variance followed by pairwise post hoc comparisons. Further, the subgroup of PD patients carrying the risky MAPT H1/H1 haplotype was compared with noncarriers (MAPT H1/H2 haplotype) in terms of CBF by a two-sample t test. A pattern that could be summarized as “posterior hypoperfusion” (PH) differentiated the PD-MCI group from the HC group with an accuracy of 92.6% (sensitivity = 93%, specificity = 93%). Additionally, the PD patients with MAPT H1/H1 haplotype had decreased perfusion than the ones with H1/H2 haplotype at the posterior areas of the visual network (VN), default mode network (DMN), and dorsal attention network (DAN). The PH-type pattern in ASL-MRI could be employed as a biomarker of both current cognitive impairment and future cognitive decline in PD.

This study aims to specify biomarkers of Parkinson's disease mild cognitive impairment (PD-MCI) based on fractional anisotropy (FA) and mean diffusivity (MD) maps obtained from diffusion weighted magnetic resonance imaging (DWMRI). T1 and diffusion weighted MR images collected from 27 cognitively normal Parkinson's disease (PD-CN), 32 mild cognitively impaired Parkinson's disease (PD-MCI), and 18 healthy control (HC) volunteers, at a clinical 3T MR scanner, were processed using FMRIB Software Library (FSL)'s toolboxes. Average regional values of FA and MD maps and tract based spatial statistics (TBSS) were utilized to define statistically significant differences between the participant subject groups.

Proton magnetic resonance spectroscopic imaging ( $$^{1}$$ H-MRSI) provides noninvasive information regarding metabolic activity within the tissues. One of the main problems of $$^{1}$$ H-MRSI is low spatial resolution due to clinical scan time limitations. Advanced post-processsing algorithms, like convolutional neural networks (CNN) might help with generation of super resolution $$^{1}$$ H-MRSI. In this study, the application of super resolution convolutional neural networks (SRCNN) for increasing the spatial resolution of $$^{1}$$ H-MRSI is presented. Fluid Attenuated Inversion Recovery (FLAIR), T1-weighted, T2-weighted magnetic resonance imaging (MRI) data and a fused MRI, which contained the three different structural MR images in each RGB channel, were used in training the SRCNN scheme. The spatial resolution of $$^{1}$$ H-MRSI images were increased by a factor of three using the models trained with the anatomical MR images. The results of the proposed technique were compared with bicubic resampling in terms of peak signal to noise ratio and root mean square error. Our results indicated that SRCNN would contribute to reconstructing higher resolution $$^{1}$$ H-MRSI.

ObjectiveTo identify the effects of oral and transdermal hormone replacement therapies (HRT) on levels of a cardiovascular disease (CVD) marker, MCP-1.DesignRandomized controlled trial.Materials and MethodsNinety nine healthy early postmenopausal women were enrolled in the study. Patients were randomly assigned to receive oral or transdermal HRT for 6 months. The first group received continuous combined oral HRT containing 1 mg 17β-estradiol and 0.5 mg norethisterone acetate (n=39), whereas the second group received sequential HRT containing estradiol alone (4 mg) given twice a week on days 1-14 and estradiol (10 mg) plus norethisterone acetate (30 mg) given twice a week on days 15-28 (n=37). Twenty patients in the oral HRT group and 19 patients in the transdermal HRT group completed the study. Circulating levels of MCP-1 were assessed before and after the treatment in all patients.ResultsTabled 1Effects of oral and transdermal HRT on MCP-1 levels, after 6 months of treatmentBaseline6 monthsP valueOral HRT (pg/ml) (n=20)150.1±12.8153.6±12.5.192Transdermal HRT (pg/ml) (n=19)145.2±11.6146.1±15.1.419HRT; hormone replacement therapy, MCP-1; monocyte chemoattractant protein-1.Moreover there was no significant difference in MCP-1 serum levels after 6 months HRT use between the groups. Open table in a new tab ConclusionBoth oral continuous and sequential transdermal HRTs do not affect MCP-1 levels significantly, which is a significant marker for endothelial damage, in young women at early postmenopausal period. ObjectiveTo identify the effects of oral and transdermal hormone replacement therapies (HRT) on levels of a cardiovascular disease (CVD) marker, MCP-1. To identify the effects of oral and transdermal hormone replacement therapies (HRT) on levels of a cardiovascular disease (CVD) marker, MCP-1. DesignRandomized controlled trial. Randomized controlled trial. Materials and MethodsNinety nine healthy early postmenopausal women were enrolled in the study. Patients were randomly assigned to receive oral or transdermal HRT for 6 months. The first group received continuous combined oral HRT containing 1 mg 17β-estradiol and 0.5 mg norethisterone acetate (n=39), whereas the second group received sequential HRT containing estradiol alone (4 mg) given twice a week on days 1-14 and estradiol (10 mg) plus norethisterone acetate (30 mg) given twice a week on days 15-28 (n=37). Twenty patients in the oral HRT group and 19 patients in the transdermal HRT group completed the study. Circulating levels of MCP-1 were assessed before and after the treatment in all patients. Ninety nine healthy early postmenopausal women were enrolled in the study. Patients were randomly assigned to receive oral or transdermal HRT for 6 months. The first group received continuous combined oral HRT containing 1 mg 17β-estradiol and 0.5 mg norethisterone acetate (n=39), whereas the second group received sequential HRT containing estradiol alone (4 mg) given twice a week on days 1-14 and estradiol (10 mg) plus norethisterone acetate (30 mg) given twice a week on days 15-28 (n=37). Twenty patients in the oral HRT group and 19 patients in the transdermal HRT group completed the study. Circulating levels of MCP-1 were assessed before and after the treatment in all patients. ResultsTabled 1Effects of oral and transdermal HRT on MCP-1 levels, after 6 months of treatmentBaseline6 monthsP valueOral HRT (pg/ml) (n=20)150.1±12.8153.6±12.5.192Transdermal HRT (pg/ml) (n=19)145.2±11.6146.1±15.1.419HRT; hormone replacement therapy, MCP-1; monocyte chemoattractant protein-1.Moreover there was no significant difference in MCP-1 serum levels after 6 months HRT use between the groups. Open table in a new tab HRT; hormone replacement therapy, MCP-1; monocyte chemoattractant protein-1. Moreover there was no significant difference in MCP-1 serum levels after 6 months HRT use between the groups. ConclusionBoth oral continuous and sequential transdermal HRTs do not affect MCP-1 levels significantly, which is a significant marker for endothelial damage, in young women at early postmenopausal period. Both oral continuous and sequential transdermal HRTs do not affect MCP-1 levels significantly, which is a significant marker for endothelial damage, in young women at early postmenopausal period.

Automatic Gestational Age (GA) estimation based on the Crown Rump Length (CRL) measurement is the preferred solution to overcome the challenges while using the last menstrual period (LMP) to date pregnancies. However, GA estimation based on CRL requires accurate placement of calipers on the fetal crown and rump which is not always a straightforward task, especially for an inexperienced sonographer. This paper proposes an accurate GA estimation method from fetal CRL images during the first trimester scan. The method addresses this problem by segmenting the fetus using a binary and multi-class U-Net. The fetal segmentation is used to compute the CRL. This is then followed by an estimation of GA from the automatic CRL measurement based of clinical information. The results from the multi-class segmentation achieves a more accurate precision, recall, Dice, and Jaccard. This has also led to a more accurate CRL measurement and hence more robust GA estimation.