Transfer Learning-Based B-Line Assessment of Lung Ultrasound for Acute Heart Failure.

OBJECTIVE: B-lines assessed by lung ultrasound (LUS) outperform physical exam, chest radiograph, and biomarkers for the associated diagnosis of acute heart failure (AHF) in the emergent setting. The use of LUS is however limited to trained professionals and suffers from interpretation variability. The objective was to utilize transfer learning to create an AI-enabled software that can aid novice users to automate LUS B-line interpretation. METHODS: Data from an observational AHF LUS study provided standardized cine clips for AI model development and evaluation. A total of 49,952 LUS frames from 30 patients were hand scored and trained on a convolutional neural network (CNN) to interpret B-lines at the frame level. A random independent evaluation set of 476 LUS clips from 60 unique patients assessed model performance. The AI models scored the clips on both a binary and ordinal 0-4 multiclass assessment. RESULTS: A multiclassification AI algorithm had the best performance at the binary level when applied to the independent evaluation set, AUC of 0.967 (95% CI 0.965-0.970) for detecting pathologic conditions. When compared to expert blinded reviewer, the 0-4 multiclassification AI algorithm scale had a reported linear weighted kappa of 0.839 (95% CI 0.804-0.871). CONCLUSIONS: The multiclassification AI algorithm is a robust and well performing model at both binary and ordinal multiclass B-line evaluation. This algorithm has the potential to be integrated into clinical workflows to assist users with quantitative and objective B-line assessment for evaluation of AHF.

Liver Biliary Function Evaluation on a 1.5T Magnetic Resonance Imaging Scan by T1 Reduction Rate Assessment Using Variable-Flip-Angle Sequences.

OBJECTIVE: Magnetic resonance (MR) relaxometry is an absolute and reproducible quantitative method, compared with signal intensity for the evaluation of liver biliary function. This is obtainable by the T1 reduction rate (T1RR), as it carries a smaller systematic error than the pre/post contrast agent T1 measurement. We aimed to develop and test an MR T1 relaxometry tool tailored for the evaluation of liver T1RR after gadolinium ethoxybenzyl-diethylenetriaminepentaacetic acid administration on 1.5T MR. METHODS: In vitro/vivo (liver) T1RR values with two 3D FLASH variable-flip-angle sequences were calculated by a MATLAB algorithm. In vitro measurements were done by 2 physicists, in consensus. The prospective in vivo study was approved by the local ethical committee and performed on 13 normal/26 cirrhotic livers. A supplemental test in 5 normal/5 cirrhotic livers, out of the studied series, was done to compare the results of our method (without B1 inhomogeneity correction) and those of a standardized commercial tool (with B1 inhomogeneity correction). All in vivo evaluations were performed by 2 radiologists with 7 years of experience in abdominal imaging. Open-source Java-based software ImageJ was used to draw the free-hand regions of interest on liver section and for the measurement of hepatic T1RR values. The T1RR values of each group of patients were compared to assess statistically significant differences. All statistical analyses were performed with IBM-SPSS Statistics. In vivo evaluations, the intrareader and interreader reliability was assessed by intraclass correlation coefficient. RESULTS: Our method showed good accuracy in evaluating in vitro T1RR with a maximum percentage error of 9% (constant at various time points) with T1 values in the 200- to 1400-millisecond range. In vivo, a high concordance between the T1RR evaluated with the proposed method and that calculated from the standardized commercial software was verified ( P < 0.05). The median T1RRs were 74.8, 67.9, and 52.1 for the normal liver, Child-Pugh A, and Child-Pugh B cirrhotic groups, respectively. A very good agreement was found, both within intrareader and interreader reliability, with intraclass correlation coefficient values ranging from 0.88 to 0.95 and from 0.85 to 0.90, respectively. CONCLUSIONS: The proposed method allowed accurate reliable in vitro/vivo T1RR assessment evaluation of the liver biliary function after gadolinium ethoxybenzyl-diethylenetriaminepentaacetic acid administration.

Automated Assessment of the DWI-FLAIR Mismatch in Patients with Acute Ischemic Stroke: Added Value to Routine Clinical Practice.

BACKGROUND AND PURPOSE: The DWI-FLAIR mismatch is used to determine thrombolytic eligibility in patients with acute ischemic stroke when the time since stroke onset is unknown. Commercial software packages have been developed for automated DWI-FLAIR classification. We aimed to use e-Stroke software for automated classification of the DWI-FLAIR mismatch in a cohort of patients with acute ischemic stroke and in a comparative analysis with 2 expert neuroradiologists. MATERIALS AND METHODS: In this retrospective study, patients with acute ischemic stroke who had MR imaging and known time since stroke onset were included. The DWI-FLAIR mismatch was evaluated by 2 neuroradiologists blinded to the time since stroke onset and automatically by the e-Stroke software. After 4 weeks, the neuroradiologists re-evaluated the MR images, this time equipped with automated predicted e-Stroke results as a computer-assisted tool. Diagnostic performances of e-Stroke software and the neuroradiologists were evaluated for prediction of DWI-FLAIR mismatch status. RESULTS: A total of 157 patients met the inclusion criteria. A total of 82 patients (52%) had a time since stroke onset of ≤4.5 hours. By means of consensus reads, 81 patients (51.5%) had a DWI-FLAIR mismatch. The diagnostic accuracy (area under the curve/sensitivity/specificity) of e-Stroke software for the determination of the DWI-FLAIR mismatch was 0.72/90.0/53.9. The diagnostic accuracy (area under the curve/sensitivity/specificity) for neuroradiologists 1 and 2 was 0.76/69.1/84.2 and 0.82/91.4/73.7, respectively; both significantly (P < .05) improved to 0.83/79.0/86.8 and 0.89/92.6/85.5, respectively, following the use of e-Stroke predictions as a computer-assisted tool. The interrater agreement (κ) for determination of DWI-FLAIR status was improved from 0.49 to 0.57 following the use of the computer-assisted tool. CONCLUSIONS: This automated quantitative approach for DWI-FLAIR mismatch provides results comparable with those of human experts and can improve the diagnostic accuracies of expert neuroradiologists in the determination of DWI-FLAIR status.

Feature-Tracking Strain Parameters Differ Between Highly Accelerated and Conventional Acquisitions: A Multisoftware Assessment.

BACKGROUND: Cardiac magnetic resonance imaging protocols have been adapted to fit the needs for faster, more efficient acquisitions, resulting in the development of highly accelerated, compressed sensing-based (CS) sequences. The aim of this study was to evaluate intersoftware and interacquisition differences for postprocessing software applied to both CS and conventional cine sequences. MATERIALS AND METHODS: A total of 106 individuals (66 healthy volunteers, 40 patients with dilated cardiomyopathy, 51% female, 38±17 y) underwent cardiac magnetic resonance at 3T with retrospectively gated conventional cine and CS sequences. Postprocessing was performed using 2 commercially available software solutions and 1 research prototype from 3 different developers. The agreement of clinical and feature-tracking strain parameters between software solutions and acquisition types was assessed by Bland-Altmann analyses and intraclass correlation coefficients. Differences between softwares and acquisitions were assessed using Kruskal-Wallis analysis of variances. In addition, receiver operating characteristic curve-derived cutoffs were used to evaluate whether sequence-specific cutoffs influence disease classification. RESULTS: There were significant intersoftware ( P <0.002 for all except LV end-diastolic volume per body surface area) and interacquisition differences ( P <0.02 for all except end-diastolic volume per body surface area from Neosoft, left ventricular mass per body surface area from cvi42 and TrufiStrain and global circumferential strain from Neosoft). However, the intraclass correlation coefficients between acquisitions were strong-to-excellent for all parameters (all ≥0.81). In comparing individual softwares to a pooled mean, Bland-Altmann analyses revealed smaller magnitudes of bias for cine acquisition than for CS acquisition. In addition, the application of conventional cutoffs to CS measurements did not result in the false reclassification of patients. CONCLUSION: Significantly lower magnitudes of strain and volumetric parameters were observed in retrospectively gated CS acquisitions, despite strong-to-excellent agreement amongst software solutions and acquisition types. It remains important to be aware of the acquisition type in the context of follow-up examinations, where different cutoffs might lead to misclassifications.

Rapid Deployment of Whole Slide Imaging for Primary Diagnosis in Surgical Pathology at Stanford Medicine: Responding to Challenges of the COVID-19 Pandemic.

CONTEXT.­: Stanford Pathology began stepwise subspecialty implementation of whole slide imaging (WSI) in 2018 soon after the first US Food and Drug Administration approval. In 2020, during the COVID-19 pandemic, the Centers for Medicare & Medicaid Services waived the requirement for pathologists to perform diagnostic tests in Clinical Laboratory Improvement Amendments (CLIA)-licensed facilities. This encouraged rapid implementation of WSI across all surgical pathology subspecialties. OBJECTIVE.­: To present our experience with validation and implementation of WSI at a large academic medical center encompassing a caseload of more than 50 000 cases per year. DESIGN.­: Validation was performed independently for 3 subspecialty services with a diagnostic concordance threshold above 95%. Analysis of user experience, staffing, infrastructure, and information technology was performed after department-wide expansion. RESULTS.­: Diagnostic concordance was achieved in 96% of neuropathology cases, 100% of gynecologic pathology cases, and 98% of immunohistochemistry cases. After full implementation, 8 high-capacity scanners were operational, with whole slide images generated on greater than 2000 slides per weekday, accounting for approximately 80% of histologic slides at Stanford Medicine. Multiple modifications in workflow and information technology were needed to improve performance. Within months of full implementation, most attending pathologists and trainees had adopted WSI for primary diagnosis. CONCLUSIONS.­: WSI across all surgical subspecialities is achievable at scale at an academic medical center; however, adoption required flexibility to adjust workflows and develop tailored solutions. WSI at scale supported the health and safety of medical staff while facilitating high-quality patient care and education during COVID-19 restrictions.

Automated quality assessment of large digitised histology cohorts by artificial intelligence.

Research using whole slide images (WSIs) of histopathology slides has increased exponentially over recent years. Glass slides from retrospective cohorts, some with patient follow-up data are digitised for the development and validation of artificial intelligence (AI) tools. Such resources, therefore, become very important, with the need to ensure that their quality is of the standard necessary for downstream AI development. However, manual quality control of large cohorts of WSIs by visual assessment is unfeasible, and whilst quality control AI algorithms exist, these focus on bespoke aspects of image quality, e.g. focus, or use traditional machine-learning methods, which are unable to classify the range of potential image artefacts that should be considered. In this study, we have trained and validated a multi-task deep neural network to automate the process of quality control of a large retrospective cohort of prostate cases from which glass slides have been scanned several years after production, to determine both the usability of the images at the diagnostic level (considered in this study to be the minimal standard for research) and the common image artefacts present. Using a two-layer approach, quality overlays of WSIs were generated from a quality assessment (QA) undertaken at patch-level at [Formula: see text] magnification. From these quality overlays the slide-level quality scores were predicted and then compared to those generated by three specialist urological pathologists, with a Pearson correlation of 0.89 for overall 'usability' (at a diagnostic level), and 0.87 and 0.82 for focus and H&E staining quality scores respectively. To demonstrate its wider potential utility, we subsequently applied our QA pipeline to the TCGA prostate cancer cohort and to a colorectal cancer cohort, for comparison. Our model, designated as PathProfiler, indicates comparable predicted usability of images from the cohorts assessed (86-90% of WSIs predicted to be usable), and perhaps more significantly is able to predict WSIs that could benefit from an intervention such as re-scanning or re-staining for quality improvement. We have shown in this study that AI can be used to automate the process of quality control of large retrospective WSI cohorts to maximise their utility for research.

Assessment of lung density in pediatric patients using three-dimensional ultrashort echo-time and four-dimensional zero echo-time sequences.

PURPOSE: Lung magnetic resonance imaging (MRI) using conventional sequences is limited due to strong signal loss by susceptibility effects of aerated lung. Our aim is to assess lung signal intensity in children on ultrashort echo-time (UTE) and zero echo-time (ZTE) sequences. We hypothesize that lung signal intensity can be correlated to lung physical density. MATERIALS AND METHODS: Lung MRI was performed in 17 children with morphologically normal lungs (median age: 4.7 years, range 15 days to 17 years). Both lungs were manually segmented in UTE and ZTE images and the average signal intensities were extracted. Lung-to-background signal ratios (LBR) were compared for both sequences and between both patient groups using non-parametric tests and correlation analysis. Anatomical region-of-interest (ROI) analysis was performed for the normal cohort for assessment of the anteroposterior lung gradient. RESULTS: There was no significant difference between LBR of normal lungs using UTE and ZTE (p < 0.05). Both sequences revealed a LBR age-dependency with a high negative correlation for UTE (Rs = - 0.77; range 2.98-1.41) and ZTE (Rs = - 0.82; range 2.66-1.38)). Signal-to-noise (SNR) and contrast-to-noise ratios (CNR) were age-dependent for both sequences. SNR was higher for children up to 2 years old with 3D UTE Cones while for the rest it was higher with 4D ZTE. CNR was similar for both sequences. Posterior lung areas exhibited higher signal intensity compared to anterior ones (UTE 9.4% and ZTE 12% higher), both with high correlation coefficients (R2UTE = 0.94, R2ZTE = 0.97). CONCLUSION: The ZTE sequence can measure signal intensity similarly to UTE in pediatric patients. Both sequences reveal an age- and gravity-dependency of LBR.

Effect of data binning and frame averaging for micro-CT image acquisition on the morphometric outcome of bone repair assessment.

Despite the current advances in micro-CT analysis, the influence of some image acquisition parameters on the morphometric assessment outcome have not been fully elucidated. The aim of this study was to determine whether data binning and frame averaging affect the morphometric outcome of bone repair assessment using micro-CT. Four Wistar rats' tibiae with a surgically created bone defect were imaged with micro-CT six times each, frame averaging set to 1 and 2, and data binning set to 1, 2 and 4, for each of the averaging values. Two-way ANOVA followed by Bonferroni tests assessed the significance of frame averaging and data binning on a set of morphometric parameters assessed in the image volumes (p < 0.01). The effect of frame averaging was not significant for any of the assessed parameters. Increased data binning led to larger trabecular thickness. In contrast, smaller bone volume fraction and bone volume were found as data binning increased. Trabeculae number and trabecular separation were not influenced by any of the parameters. In conclusion, the morphometric outcome of bone repair assessment in micro-CT demonstrated dependency upon data binning, but not frame averaging. Therefore, image acquisition of small anatomical structures (e.g., rat trabeculae) should be performed without data binning.

Correlation Tensor MRI deciphers underlying kurtosis sources in stroke.

Noninvasively detecting and characterizing modulations in cellular scale micro-architecture remains a desideratum for contemporary neuroimaging. Diffusion MRI (dMRI) has become the mainstay methodology for probing microstructure, and, in ischemia, its contrasts have revolutionized stroke management. Diffusion kurtosis imaging (DKI) has been shown to significantly enhance the sensitivity of stroke detection compared to its diffusion tensor imaging (DTI) counterparts. However, the interpretation of DKI remains ambiguous as its contrast may arise from competing kurtosis sources related to the anisotropy of tissue components, diffusivity variance across components, and microscopic kurtosis (e.g., arising from cross-sectional variance, structural disorder, and restriction). Resolving these sources may be fundamental for developing more specific imaging techniques for stroke management, prognosis, and understanding its pathophysiology. In this study, we apply Correlation Tensor MRI (CTI) - a double diffusion encoding (DDE) methodology recently introduced for deciphering kurtosis sources based on the unique information captured in DDE's diffusion correlation tensors - to investigate the underpinnings of kurtosis measurements in acute ischemic lesions. Simulations for the different kurtosis sources revealed specific signatures for cross-sectional variance (representing neurite beading), edema, and cell swelling. Ex vivo CTI experiments at 16.4 T were then performed in an experimental photothrombotic stroke model 3 h post-stroke (N = 10), and successfully separated anisotropic, isotropic, and microscopic non-Gaussian diffusion sources in the ischemic lesions. Each of these kurtosis sources provided unique contrasts in the stroked area. Particularly, microscopic kurtosis was shown to be a primary "driver" of total kurtosis upon ischemia; its large increases, coupled with decreases in anisotropic kurtosis, are consistent with the expected elevation in cross-sectional variance, likely linked to beading effects in small objects such as neurites. In vivo experiments at 9.4 T at the same time point (3 h post ischemia, N = 5) demonstrated the stability and relevance of the findings and showed that fixation is not a dominant confounder in our findings. In future studies, the different CTI contrasts may be useful to address current limitations of stroke imaging, e.g., penumbra characterization, distinguishing lesion progression form tissue recovery, and elucidating pathophysiological correlates.

Embryonic echocardiography for assessment of congenital and functional cardiac defects.

Cardiac function and morphology by mouse fetal echocardiography can be assessed by scanning the uterus extracted from the abdominal cavity (trans-uterine ultrasound) or the womb (trans-abdominal ultrasound). Advantages of trans-abdominal ultrasound include (1) non-invasive longitudinal analysis at different stages, reducing animal use; and (2) maintenance of natural environment, diminishing perturbations on functional parameters, which are more frequent in trans-uterine conditions. Here we describe both approaches, explaining how to identify congenital cardiac defects and defining the correlation between echocardiography findings and histological analysis. For complete details on the use and execution of this protocol, please refer to (Menendez-Montes et al., 2016) and (Menendez-Montes et al., 2021).

Recycling diagnostic MRI for empowering brain morphometric research - Critical & practical assessment on learning-based image super-resolution.

Preliminary studies have shown the feasibility of deep learning (DL)-based super-resolution (SR) technique for reconstructing thick-slice/gap diagnostic MR images into high-resolution isotropic data, which would be of great significance for brain research field if the vast amount of diagnostic MRI data could be successively put into brain morphometric study. However, less evidence has addressed the practicability of the strategy, because lack of a large-sample available real data for constructing DL model. In this work, we employed a large cohort (n = 2052) of peculiar data with both low through-plane resolution diagnostic and high-resolution isotropic brain MR images from identical subjects. By leveraging a series of SR approaches, including a proposed novel DL algorithm of Structure Constrained Super Resolution Network (SCSRN), the diagnostic images were transformed to high-resolution isotropic data to meet the criteria of brain research in voxel-based and surface-based morphometric analyses. We comprehensively assessed image quality and the practicability of the reconstructed data in a variety of morphometric analysis scenarios. We further compared the performance of SR approaches to the ground truth high-resolution isotropic data. The results showed (i) DL-based SR algorithms generally improve the quality of diagnostic images and render morphometric analysis more accurate, especially, with the most superior performance of the novel approach of SCSRN. (ii) Accuracies vary across brain structures and methods, and (iii) performance increases were higher for voxel than for surface based approaches. This study supports that DL-based image super-resolution potentially recycle huge amount of routine diagnostic brain MRI deposited in sleeping state, and turning them into useful data for neurometric research.

Pain assessment in horses using automatic facial expression recognition through deep learning-based modeling.

The aim of this study was to develop and evaluate a machine vision algorithm to assess the pain level in horses, using an automatic computational classifier based on the Horse Grimace Scale (HGS) and trained by machine learning method. The use of the Horse Grimace Scale is dependent on a human observer, who most of the time does not have availability to evaluate the animal for long periods and must also be well trained in order to apply the evaluation system correctly. In addition, even with adequate training, the presence of an unknown person near an animal in pain can result in behavioral changes, making the evaluation more complex. As a possible solution, the automatic video-imaging system will be able to monitor pain responses in horses more accurately and in real-time, and thus allow an earlier diagnosis and more efficient treatment for the affected animals. This study is based on assessment of facial expressions of 7 horses that underwent castration, collected through a video system positioned on the top of the feeder station, capturing images at 4 distinct timepoints daily for two days before and four days after surgical castration. A labeling process was applied to build a pain facial image database and machine learning methods were used to train the computational pain classifier. The machine vision algorithm was developed through the training of a Convolutional Neural Network (CNN) that resulted in an overall accuracy of 75.8% while classifying pain on three levels: not present, moderately present, and obviously present. While classifying between two categories (pain not present and pain present) the overall accuracy reached 88.3%. Although there are some improvements to be made in order to use the system in a daily routine, the model appears promising and capable of measuring pain on images of horses automatically through facial expressions, collected from video images.

A pyramidal deep learning pipeline for kidney whole-slide histology images classification.

Renal cell carcinoma is the most common type of kidney cancer. There are several subtypes of renal cell carcinoma with distinct clinicopathologic features. Among the subtypes, clear cell renal cell carcinoma is the most common and tends to portend poor prognosis. In contrast, clear cell papillary renal cell carcinoma has an excellent prognosis. These two subtypes are primarily classified based on the histopathologic features. However, a subset of cases can a have a significant degree of histopathologic overlap. In cases with ambiguous histologic features, the correct diagnosis is dependent on the pathologist's experience and usage of immunohistochemistry. We propose a new method to address this diagnostic task based on a deep learning pipeline for automated classification. The model can detect tumor and non-tumoral portions of kidney and classify the tumor as either clear cell renal cell carcinoma or clear cell papillary renal cell carcinoma. Our framework consists of three convolutional neural networks and the whole slide images of kidney which were divided into patches of three different sizes for input into the networks. Our approach can provide patchwise and pixelwise classification. The kidney histology images consist of 64 whole slide images. Our framework results in an image map that classifies the slide image on the pixel-level. Furthermore, we applied generalized Gauss-Markov random field smoothing to maintain consistency in the map. Our approach classified the four classes accurately and surpassed other state-of-the-art methods, such as ResNet (pixel accuracy: 0.89 Resnet18, 0.92 proposed). We conclude that deep learning has the potential to augment the pathologist's capabilities by providing automated classification for histopathological images.

Development of a multiplex immuno-oncology biomarker and digital pathology workflow for assessment of urothelial carcinoma.

BACKGROUND: Immune checkpoint inhibitors (ICI) therapies have demonstrated significant benefit in the treatment of many tumors including high grade urothelial cancer (HGUC) of the bladder. However, variability in patients' clinical responses highlights the need for biomarkers to aid patient stratification. ICI relies on an intact host immune response. In this context, we hypothesize that key players in the antitumor immune response such as markers of activated cytotoxic T lymphocytes (CD8, granzyme-B) and immune suppression (FOXP3) may help to identify patients who will derive the greatest therapeutic benefit from ICI. A major obstacle for deployment of such a strategy is the limited quantities of tumor-derived biopsy material. Therefore, in this technical study, we develop a multiplex biomarker with digital workflow. We explored the (1) concordance of conventional single stain results using digital image analysis, and (2) agreement between digital scoring versus manual analysis. METHODS: (1) For concordance study of single and multiplex stains, triplicate core tissue microarrays of 207 muscle invasive, HGUC of bladder had sequential 4-micron sections cut and stained with CD8, FOXP3 and granzyme-B. An inhouse developed tri-chromogen multiplex immunohistochemistry (m-IHC) assay consisting of CD8 (green), granzyme B (brown), and FOXP3 (red) was used to stain the next sequential tissue section. (2) Agreement between manual and digital analysis was performed on 19 whole slide sections of HGUC cystectomy specimens. All slides were scanned using Aperio ScanScope AT Digital Scanner at 40X. Quantitative digital image analysis was performed using QuPath version 0.2.3 open-source software. Scores from triplicate cores were averaged for each HGUC specimen for each marker. Intraclass correlation coefficients were used to compare percent positive cells between the single- and multi-plex assays. Lin's concordance correlation coefficients were used for manual versus digital analysis. RESULTS AND CONCLUSIONS: m-IHC offers significant advantages in characterizing the host immune microenvironment particularly in limited biopsy tissue material. Utilizing a digital image workflow resulted in significant concordance between m-IHC and individual single stains (p < 0.001 for all assessments). Moderate to good agreements were achieved between manual and digital scoring. Our technical work demonstrated potential uses of multiplex marker in assessing the host immune status and could be used in conjunction with PD-L1 as a predictor of response to ICI therapy.

Optimization of psoriasis assessment system based on patch images.

Psoriasis is a chronic inflammatory skin disease that occurs in various forms throughout the body and is associated with certain conditions such as heart disease, diabetes, and depression. The psoriasis area severity index (PASI) score, a tool used to evaluate the severity of psoriasis, is currently used in clinical trials and clinical research. The determination of severity is based on the subjective judgment of the clinician. Thus, the disease evaluation deviations are induced. Therefore, we propose optimal algorithms that can effectively segment the lesion area and classify the severity. In addition, a new dataset on psoriasis was built, including patch images of erythema and scaling. We performed psoriasis lesion segmentation and classified the disease severity. In addition, we evaluated the best-performing segmentation method and classifier and analyzed features that are highly related to the severity of psoriasis. In conclusion, we presented the optimal techniques for evaluating the severity of psoriasis. Our newly constructed dataset improved the generalization performance of psoriasis diagnosis and evaluation. It proposed an optimal system for specific evaluation indicators of the disease and a quantitative PASI scoring method. The proposed system can help to evaluate the severity of localized psoriasis more accurately.

The utility of digital pathology in improving the diagnostic skills of pathology trainees in commonly encountered pigmented cutaneous lesions during the COVID-19 pandemic: A single academic institution experience.

Digital pathology has become an integral part of pathology education in recent years, particularly during the COVID-19 pandemic, for its potential utility as a teaching tool that augments the traditional 1-to-1 sign-out experience. Herein, we evaluate the utility of whole slide imaging (WSI) in reducing diagnostic errors in pigmented cutaneous lesions by pathology fellows without subspecialty training in dermatopathology. Ten cases of 4 pigmented cutaneous lesions commonly encountered by general pathologists were selected. Corresponding whole slide images were distributed to our fellows, along with two sets of online surveys, each composed of 10 multiple-choice questions with 4 answers. Identical cases were used for both surveys to minimize variability in trainees' scores depending on the perceived level of difficulty, with the second set being distributed after random shuffling. Brief image-based teaching slides as self-assessment tool were provided to trainees between each survey. Pre- and post-self-assessment scores were analyzed. 61% (17/28) and 39% (11/28) of fellows completed the first and second surveys, respectively. The mean score in the first survey was 5.2/10. The mean score in the second survey following self-assessment increased to 7.2/10. 64% (7/11) of trainees showed an improvement in their scores, with 1 trainee improving his/her score by 8 points. No fellow scored less post-self-assessment than on the initial assessment. The difference in individual scores between two surveys was statistically significant (p = 0.003). Our study demonstrates the utility of WSI-based self-assessment learning as a source of improving diagnostic skills of pathology trainees in a short period of time.

Visual assessment of mitotic figures in breast cancer: a comparative study between light microscopy and whole slide images.

BACKGROUND AND AIMS: Visual assessment of mitotic figures in breast cancer (BC) remains a challenge. This is expected to be more pronounced in the digital pathology era. This study aims to refine the criteria of mitotic figure recognition, particularly in whole slide images (WSI). METHOD AND RESULTS: Haematoxylin and eosin (H&E)-stained BC sections (n = 506) were examined using light microscopy (LM) and WSI. A set of features for identifying mitosis in WSI and to distinguish true figures from mimickers was developed. Changes in the mitotic count between the two platforms was explored. Morphological features of mitoses were recorded separately, including absence of nuclear membrane, chromatin hairy-like projections, shape, cytoplasmic features, mitotic cell size and relationship to surrounding cells. Each mitotic phase has its own mimickers. Fifty-eight per cent of mitoses showed absent hairy-like projection in WSI; however, 89% retained their ragged nuclear border, which distinguished them from mimickers including apoptotic cells, lymphocytes and dark elongated hyperchromatic structures. Mitosis in WSI showed loss of fine details, and there was a 20% average reduction rate of mitotic counts when compared to the same area on LM. Using refined mitosis recognition criteria in WSI resulted in a twofold improvement of interobserver concordance. However, when compared to LM, 19% of cases were underscored in WSIs. CONCLUSIONS: All morphological features of mitosis should be considered to enable recognition and differentiation from their mimickers, particularly in WSI, to ensure reliable BC grading. Refining mitotic cut-offs per specific area when using WSI, based on the degree of reduction and association with outcome, is warranted.

Radiographs assessment of changes in the psoas muscle at L4-L5 level after single-level lateral lumbar interbody fusion in patients with postoperative motor weakness.

The purposes of this study were (1) to investigate postoperative changes in cross-sectional area (CSA) and signal intensity (SI) of the psoas muscle (PS) using magnetic resonance imaging (MRI) and (2) to compare the CSA and SI of the PS between patients with and without motor weakness after single-level lateral lumbar interbody fusion (LLIF) at level L4-L5. Sixty patients were divided into two groups-those with postoperative motor weakness and those without-and the two groups were compared. Baseline demographics and clinical characteristics, such as operation time and blood loss, length of hospital stay, and postoperative complications, were recorded. The CSA and SI of the PS were obtained from the MRI regions of interest defined by manual tracing. Patients who developed motor weakness after surgery were significantly older (p = 0.040). The operation time (p = 0.868), LLIF operative time (p = 0.476), and estimated bleeding loss (p = 0.168) did not differ significantly between groups. In both groups, the CSA and SI of the left and right PS increased after surgery. The change in the CSA of the left PS was significantly higher in patients with weakness (247.6 ± 155.2 mm2) than without weakness (152.2 ± 133.1 mm2) (p = 0.036). The change in SI of the left PS did not differ between the two groups (p = 0.530). To prevent postoperative motor weakness regardless of the operation time, surgeons should be aware of the potential for surgical invasive of the PS during LLIF in older people.

Impact of fully automated assessment on interstudy reproducibility of biventricular volumes and function in cardiac magnetic resonance imaging.

Cardiovascular magnetic resonance (CMR) imaging provides reliable assessments of biventricular morphology and function. Since manual post-processing is time-consuming and prone to observer variability, efforts have been directed towards novel artificial intelligence-based fully automated analyses. Hence, we sought to investigate the impact of artificial intelligence-based fully automated assessments on the inter-study variability of biventricular volumes and function. Eighteen participants (11 with normal, 3 with heart failure and preserved and 4 with reduced ejection fraction (EF)) underwent serial CMR imaging at in median 63 days (range 49-87) interval. Short axis cine stacks were acquired for the evaluation of left ventricular (LV) mass, LV and right ventricular (RV) end-diastolic, end-systolic and stroke volumes as well as EF. Assessments were performed manually (QMass, Medis Medical Imaging Systems, Leiden, Netherlands) by an experienced (3 years) and inexperienced reader (no active reporting, 45 min of training with five cases from the SCMR consensus data) as well as fully automated (suiteHEART, Neosoft, Pewaukee, WI, USA) without any manual corrections. Inter-study reproducibility was overall excellent with respect to LV volumetric indices, best for the experienced observer (intraclass correlation coefficient (ICC) > 0.98, coefficient of variation (CoV, < 9.6%) closely followed by automated analyses (ICC > 0.93, CoV < 12.4%) and lowest for the inexperienced observer (ICC > 0.86, CoV < 18.8%). Inter-study reproducibility of RV volumes was excellent for the experienced observer (ICC > 0.88, CoV < 10.7%) but considerably lower for automated and inexperienced manual analyses (ICC > 0.69 and > 0.46, CoV < 22.8% and < 28.7% respectively). In this cohort, fully automated analyses allowed reliable serial investigations of LV volumes with comparable inter-study reproducibility to manual analyses performed by an experienced CMR observer. In contrast, RV automated quantification with current algorithms still relied on manual post-processing for reliability.

Automatic Segmentation of Left Ventricle in Echocardiography Based on YOLOv3 Model to Achieve Constraint and Positioning.

Cardiovascular disease (CVD) is the most common type of disease and has a high fatality rate in humans. Early diagnosis is critical for the prognosis of CVD. Before using myocardial tissue strain, strain rate, and other indicators to evaluate and analyze cardiac function, accurate segmentation of the left ventricle (LV) endocardium is vital for ensuring the accuracy of subsequent diagnosis. For accurate segmentation of the LV endocardium, this paper proposes the extraction of the LV region features based on the YOLOv3 model to locate the positions of the apex and bottom of the LV, as well as that of the LV region; thereafter, the subimages of the LV can be obtained, and based on the Markov random field (MRF) model, preliminary identification and binarization of the myocardium of the LV subimages can be realized. Finally, under the constraints of the three aforementioned positions of the LV, precise segmentation and extraction of the LV endocardium can be achieved using nonlinear least-squares curve fitting and edge approximation. The experiments show that the proposed segmentation evaluation indices of the method, including computation speed (fps), Dice, mean absolute distance (MAD), and Hausdorff distance (HD), can reach 2.1-2.25 fps, 93.57 ± 1.97%, 2.57 ± 0.89 mm, and 6.68 ± 1.78 mm, respectively. This indicates that the suggested method has better segmentation accuracy and robustness than existing techniques.