Stain-Independent Deep Learning-Based Analysis of Digital Kidney Histopathology.

Convolutional neural network (CNN)-based image analysis applications in digital pathology (eg, tissue segmentation) require a large amount of annotated data and are mostly trained and applicable on a single stain. Here, a novel concept based on stain augmentation is proposed to develop stain-independent CNNs requiring only one annotated stain. In this benchmark study on stain independence in digital pathology, this approach is comprehensively compared with state-of-the-art techniques including image registration and stain translation, and several modifications thereof. A previously developed CNN for segmentation of periodic acid-Schiff-stained kidney histology was used and applied to various immunohistochemical stainings. Stain augmentation showed very high performance in all evaluated stains and outperformed all other techniques in all structures and stains. Without the need for additional annotations, it enabled segmentation on immunohistochemical stainings with performance nearly comparable to that of the annotated periodic acid-Schiff stain and could further uphold performance on several held-out stains not seen during training. Herein, examples of how this framework can be applied for compartment-specific quantification of immunohistochemical stains for inflammation and fibrosis in animal models and patient biopsy specimens are presented. The results show that stain augmentation is a highly effective approach to enable stain-independent applications of deep-learning segmentation algorithms. This opens new possibilities for broad implementation in digital pathology.

Comprehensive clinicopathologic study of alpha fetoprotein-expression in a large cohort of patients with hepatocellular carcinoma.

Alpha fetoprotein (AFP) is the most widely used diagnostic and prognostic serum biomarker for hepatocellular carcinoma (HCC). Despite its wide clinical use, a systematic clinicopathologic study comparing AFP expression in HCC in situ with serum AFP concentrations has not yet been conducted. To analyze AFP expression in a large cohort of patients by immunohistochemistry, we employed a comprehensive tissue microarray with 871 different HCCs of overall 561 patients. AFP immunoreactivity was detected in only about 20% of HCC core biopsies, whereas 48.9% of the patients displayed increased serum values (>12 ng/mL). Immunostaining of whole tumor slides revealed that lack of detectable immunoreactivity in core biopsies in a subgroup of patients with elevated AFP serum concentrations is due to heterogeneous intratumoral AFP expression. Serum AFP concentrations and AFP expression in situ were moderately correlated (Spearman's rank correlation coefficient .53, P = 1.2e - 13). High AFP expression detected in serum (>227.3 ng/mL) or in situ predicted unfavorable prognosis and was associated with vascular invasion, higher tumor grade and macrotrabecular-massive tumor subtype. Multivariate and ROC curve analysis demonstrated that high AFP concentrations in serum is an independent prognostic parameter and represents the more robust prognostic predictor in comparison to AFP immunostaining of core biopsies. The previously published vessels encapsulating tumor clusters (VETC) pattern turned out as an additional, statistically independent prognostic parameter. AFP-positivity was associated with increased tumor cell apoptosis, but not with increased vascular densities. Additionally, AFP-positive tumors displayed increased proliferation rates, urea cycle dysregulation and signs of genomic instability, which may constitute the basis for their increased aggressiveness.

A novel point-of-care device accurately measures thyrotropin in whole blood, capillary blood and serum.

OBJECTIVES: Point-of-care (POC) measurement of thyrotropin (TSH) may facilitate prompt diagnosis of thyroid dysfunction. We evaluated the analytical performance of a new POC TSH assay (Wondfo). METHODS: TSH measurements were made from 730 consecutive, unselected subjects in an outpatient setting, using Wondfo in whole blood, capillary blood and serum or automated reference equipment (serum only). RESULTS: TSH measurements were user-independent. Total intra-and inter-assay variation (CV%) was 12.1 and 16.2%, respectively. Total CV% was 10.6-22.6% and 14.5-21.6% in serum and whole blood, respectively. Linearity was very good. Recovery rate was 97-127%. Prolongation of incubation time increased TSH results of 12% (13%) and 33% (35%) after 2 and 5 additional minutes in serum (blood), respectively. When measured simultaneously in two Wondfo devices, the slope of the regression line was 1.03 (serum) and 1.02 (blood), with Spearman's correlation of 0.99 for both. TSH measurements between Wondfo and reference correlated strongly (r=0.93-0.96), though TSH measurements were lower with Wondfo (slopes of plots of measurements made using the two devices were 0.94 [serum vs. serum]; 0.83 [whole blood vs. serum] and 0.64 [capillary blood vs. serum]). Depending on sample material, TSH in capillary blood was lower vs. whole blood (slope: 0.82) and for whole blood vs. serum (Wondfo and reference method; slope: 0.69 and 0.83). Total haemolysis, but not elevated bilirubin or lipemia, disrupted TSH measurement. CONCLUSIONS: The Wondfo system was straightforward to use without need for specialist technicians and demonstrated analytic performance suitable for clinical use for the diagnosis of thyroid dysfunction.

Multi-center evaluation of the highly sensitive Abbott ARCHITECT and Alinity thyroglobulin chemiluminescent microparticle immunoassay.

BACKGROUND: Thyroglobulin (Tg) is an essential part for the management of patients with differentiated thyroid carcinoma (DTC) after thyroidectomy. Highly sensitive Tg assays are now established in clinical practice as they facilitate follow-up of DTC patients. In this study, we evaluated the recently launched highly sensitive Abbott Tg assay for Alinity and ARCHITECT. METHODS: In this three-center study, Tg values of 447 routine patient samples, characterized for the presence of anti-Tg, were measured with the ARCHITECT Tg assay and compared with the Roche Elecsys TgII assay. In addition, a subset of 154 samples was compared also with the Beckman Tg Access assay and another subset (n = 122) with Abbott Tg on the Alinity i analyzer. RESULTS: LoQ was verified to be less than or equal to 0.1 ng/ml confirming that the Tg assay on ARCHITECT and Alinity is highly sensitive. Correlation of ARCHITECT, Alinity, and Roche was excellent with a slope between 0.9 and 1.1 and a correlation coefficient >0.98. Correlation to Beckmann Tg was also very good, but the differences in absolute values were significant (slope: 1.477). CONCLUSIONS: The Abbott Thyroglobulin assay, which is standardized to CRM-457, demonstrated a high correlation to the Roche and Beckman Tg assays, though good agreement of absolute values was only observed between Abbott and Roche. Strength of correlation and slope were not affected by the presence of anti-Tg indicating that all assays included in the study have a similar susceptibility to anti-Tg.

Neurofilament light chain levels reflect outcome in a patient with glutamic acid decarboxylase 65 antibody-positive autoimmune encephalitis under immune checkpoint inhibitor therapy.

Neurological immune-mediated side effects are rare but often severe complications of immune checkpoint inhibitor (ICI) treatment. This report describes a severe case of nivolumab/ipilimumab-associated glutamic acid decarboxylase 65-positive autoimmune encephalitis. It proposes neurofilament light chain levels, a biomarker indicating axonal damage, in the cerebrospinal fluid and serum as a putative novel biomarker for this diagnostically and therapeutically challenging entity with an often unfavorable outcome. Additionally, we provide an overview of previous reports of patients developing autoimmune encephalitis under ICI treatment.

Distribution of HOMA-IR in a population-based cohort and proposal for reference intervals.

OBJECTIVES: Insulin resistance (IR) is a hallmark of type 2 diabetes mellitus (DM). The homeostatic model assessment of insulin resistance (HOMA-IR) provides an estimate for IR from fasting glucose and insulin serum concentrations. The aim of this study was to obtain a reference interval for HOMA-IR for a specific insulin immunoassay. METHODS: The Gutenberg Health Study (GHS) is a population-based, prospective, single-center cohort study in Germany with 15,030 participants aged 35-74 years. Fasting glucose, insulin, and C-peptide were available in 10,340 participants. HOMA-IR was calculated in this group and three reference subgroups with increasingly more stringent inclusion criteria. Age- and sex-dependent distributions of HOMA-IR and reference intervals were obtained. In a substudy three insulin assays were compared and HOMA-IR estimated for each assay. RESULTS: Among the 10,340 participants analyzed there were 6,590 non-diabetic, 2,901 prediabetic, and 849 diabetic individuals. Median (interquartile range [IQR]) HOMA-IR was 1.54 (1.13/2.19), 2.00 (1.39/2.99), and 4.00 (2.52/6.51), respectively. The most stringently selected reference group consisted of 1,065 persons. Median (IQR) HOMA-IR was 1.09 (0.85/1.42) with no significant difference between men and women. The 97.5th percentile was 2.35. There was a non-significant trend towards higher values with older age. Comparison of three immunoassays for insulin showed an unsatisfactory correlation among the assays and systematic differences in calculated HOMA-IR. CONCLUSIONS: We present HOMA-IR reference intervals for adults derived by more or less stringent selection criteria for the reference cohort. In addition we show that assay specific reference intervals for HOMA-IR are required.

[Comparison of the Accuracy of the Measurement of Fetal Hemoglobin by Blood Gas Analysis and by Laboratory Gold Standard: a Prospective Diagnostic Study]. / Prospektiver Vergleich der Genauigkeit der Messung des fetalen Hämoglobins mittels Blutgasanalyse mit dem Goldstandard bei früh- und reifgeborenen Neugeborenen.

BACKGROUND: In neonatologic clinical practice and research the percentage of fetal hemoglobin (HbF) of total hemoglobin can be of interest. Blood gas analyzers offer the measurement of HbF. However, it is not known if results are accurate enough to apply in clinical decision-making or scientific questions. In this prospective diagnostic study, we examined the accuracy of HbF measurement by a blood gas analyzer. METHODS: On a neonatal intensive care and neonatal ward, the percentage of HbF was measured using both the laboratory gold standard (HbFlab, reference method) and the blood gas analyzer (HbFgas) (ABL 800 Flex, Radiometer). Agreement of HbFlab and HbFgas was assessed by the Bland-Altman method including bias and limits of agreement and by calculation of the root mean square error (RMSE). RESULTS: Thirty-five measurements in 23 term and preterm infants with a median body weight of 2190 g (min-max 967-3800 g) and a median postmenstrual age of 36+1 weeks (min-max 29+6-43+2) were performed. The Bland-Altman diagram for the measurement of HbF(gas) versus HbF(lab) shows an overestimation of HbF by the blood gas analyzer (bias 9.3%, limits of agreement 1 to 17.6%). RMSE was 10.2%; 45.7% of HbFgas measurements were >10% out of range from HbFlab. There was no influence of age, body temperature or oxygen saturation on the bias (p=0,132; p=0,194; p=0,970), but bias increased with increasing HbFlab (Pearson correlation r=0,426; p=0,011). CONCLUSION: The measurement of HbF in term and preterm infants by a blood gas analyzer lacked sufficient agreement with that of the reference method to recommend this application for clinical decision-making or scientific purposes.

Activation of silent mating type information regulation 2 homolog 1 by human chorionic gonadotropin exerts a therapeutic effect on hepatic injury and inflammation.

Incidence and prevalence of inflammatory liver diseases has increased over the last years, but therapeutic options are limited. Pregnancy induces a state of immune tolerance, which can result in spontaneous improvement of clinical symptoms of certain autoimmune diseases including autoimmune hepatitis (AIH). We investigated the immune-suppressive mechanisms of the human pregnancy hormone, chorionic gonadotropin (hCG), in the liver. hCG signaling activates silent mating type information regulation 2 homolog 1 (SIRT1), which deacetylates forkhead box o3 (FOXO3a), leading to repression of proapoptotic gene expression, because the immunosuppressive consequence attributed to the absence of caspase-3 activity of hepatocellular interleukin 16 (IL-16) is no longer processed and released. Thus, serum levels of IL-16, a key chemotactic factor for CD4+ lymphocytes, were reduced and migration to injured hepatocytes prevented. Furthermore, elevated IL-16 levels are found in the sera from patients with AIH, hepatitis B virus, hepatitis C virus, and nonalcoholic steatohepatitis. CONCLUSION: Here, we report that hCG regulates the SIRT1/FOXO3a axis in hepatocytes, resulting in immune suppression by attenuating caspase-3-dependent IL-16 processing and release, which concomitantly prevents autoaggressive T-cell infiltration of the liver. Considering the low toxicity profile of hCG in humans, interrupting the inflammatory cycle by hCG opens new perspectives for therapeutic intervention of inflammatory liver diseases. (Hepatology 2017;65:2074-2089).

Integration of 3D multimodal imaging data of a head and neck cancer and advanced feature recognition.

In the last years, matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI MSI) became an imaging technique which has the potential to characterize complex tumor tissue. The combination with other modalities and with standard histology techniques was achieved by the use of image registration methods and enhances analysis possibilities. We analyzed an oral squamous cell carcinoma with up to 162 consecutive sections with MALDI MSI, hematoxylin and eosin (H&E) staining and immunohistochemistry (IHC) against CD31. Spatial segmentation maps of the MALDI MSI data were generated by similarity-based clustering of spectra. Next, the maps were overlaid with the H&E microscopy images and the results were interpreted by an experienced pathologist. Image registration was used to fuse both modalities and to build a three-dimensional (3D) model. To visualize structures below resolution of MALDI MSI, IHC was carried out for CD31 and results were embedded additionally. The integration of 3D MALDI MSI data with H&E and IHC images allows a correlation between histological and molecular information leading to a better understanding of the functional heterogeneity of tumors. This article is part of a Special Issue entitled: MALDI Imaging, edited by Dr. Corinna Henkel and Prof. Peter Hoffmann.

Prospective analysis of long-term renal function in survivors of childhood Wilms tumor.

BACKGROUND: Considering the improved outcome, a better understanding of the late effects in Wilms tumor survivors (WT-S) is needed. This study was aimed at evaluating renal function and determining the prevalence of clinical and subclinical renal dysfunction in a cohort of WT-S using a multimodal diagnostic approach. METHODS: Thirty-seven WT-S were included in this prospective cross-sectional single center study. To evaluate renal function, glomerular filtration rate (GFR) and urinary protein excretion were assessed. Additionally, kidney sonomorphology and blood pressure were analyzed. RESULTS: All examined WT-S (mean age 28.7 years, mean follow-up 24.8 years) had been treated with a combination of surgery and chemotherapy; 59.5% had received adjuvant radiotherapy. Impaired glomerular renal function was detected in a considerable proportion of WT-S, with age-adjusted cystatin-based GFR estimation below age norm in 55.9%. A lower cystatin-based estimated GFR (eGFR) correlated with longer follow-up time and higher irradiation dose. In 5 patients (13.5%) albuminuria was identified. Analysis of sonomorphology detected compensatory contralateral renal hypertrophy in 83.3% of WT-S. Chronic kidney disease (CKD) ≥ stage II was present in 55.9% of WT-S. Blood pressure measurements revealed arterial hypertension in 15 (40.5%) WT-S (newly diagnosed n=10). In 24.3% both CKD ≥ stage II and arterial hypertension were determined. CONCLUSION: Even though WT-S are believed to carry a low risk for end-stage renal disease, in this study, a remarkable number of WT-S presented with previously unidentified subclinical signs of renal function impairment and secondary morbidity. Therefore, it is important to continue regular follow-up, especially after transition into adulthood.

Evaluation of a next generation direct whole blood enzymatic assay for hemoglobin A1c on the ARCHITECT c8000 chemistry system.

BACKGROUND: The utility of HbA1c for the diagnosis of type 2 diabetes requires an accurate, precise and robust test measurement system. Currently, immunoassay and HPLC are the most popular methods for HbA1c quantification, noting however the limitations associated with some platforms, such as imprecision or interference from common hemoglobin variants. Abbott Diagnostics has introduced a fully automated direct enzymatic method for the quantification of HbA1c from whole blood on the ARCHITECT chemistry system. METHODS: Here we completed a method evaluation of the ARCHITECT HbA1c enzymatic assay for imprecision, accuracy, method comparison, interference from hemoglobin variants and specimen stability. This was completed at three independent clinical laboratories in North America and Europe. RESULTS: The total imprecision ranged from 0.5% to 2.2% CV with low and high level control materials. Around the diagnostic cut-off of 48 mmol/mol, the total imprecision was 0.6% CV. Mean bias using reference samples from IFCC and CAP ranged from -1.1 to 1.0 mmol/mol. The enzymatic assay also showed excellent agreement with HPLC methods, with slopes of 1.01 and correlation coefficients ranging from 0.984 to 0.996 compared to Menarini Adams HA-8160, Bio-Rad Variant II and Variant II Turbo instruments. Finally, no significant effect was observed for erythrocyte sedimentation or interference from common hemoglobin variants in patient samples containing heterozygous HbS, HbC, HbD, HbE, and up to 10% HbF. CONCLUSIONS: The ARCHITECT enzymatic assay for HbA1c is a robust and fully automated method that meets the performance requirements to support the diagnosis of type 2 diabetes.

European multicenter analytical evaluation of the Abbott ARCHITECT STAT high sensitive troponin I immunoassay.

BACKGROUND: International recommendations highlight the superior value of cardiac troponins (cTns) for early diagnosis of myocardial infarction along with analytical requirements of improved precision and detectability. In this multicenter study, we investigated the analytical performance of a new high sensitive cardiac troponin I (hs-cTnI) assay and its 99th percentile upper reference limit (URL). METHODS: Laboratories from nine European countries evaluated the ARCHITECT STAT high sensitive troponin I (hs-TnI) immunoassay on the ARCHITECT i2000SR/i1000SR immunoanalyzers. Imprecision, limit of blank (LoB), limit of detection (LoD), limit of quantitation (LoQ) linearity of dilution, interferences, sample type, method comparisons, and 99th percentile URLs were evaluated in this study. RESULTS: Total imprecision of 3.3%-8.9%, 2.0%-3.5% and 1.5%-5.2% was determined for the low, medium and high controls, respectively. The lowest cTnI concentration corresponding to a total CV of 10% was 5.6 ng/L. Common interferences, sample dilution and carryover did not affect the hs-cTnI results. Slight, but statistically significant, differences with sample type were found. Concordance between the investigated hs-cTnI assay and contemporary cTnI assay at 99th percentile cut-off was found to be 95%. TnI was detectable in 75% and 57% of the apparently healthy population using the lower (1.1 ng/L) and upper (1.9 ng/L) limit of the LoD range provided by the ARCHITECT STAT hs-TnI package insert, respectively. The 99th percentile values were gender dependent. CONCLUSIONS: The new ARCHITECT STAT hs-TnI assay with improved analytical features meets the criteria of high sensitive Tn test and will be a valuable diagnostic tool.

Anti-NMDA-receptor encephalitis and concurrent neuroborreliosis misdiagnosed for post-COVID-19-syndrome: a case report.

We present a case of a 42-year-old woman with paraneoplastic anti-N-Methyl-D-Aspartat (NMDA)-receptor encephalitis and concurrent neuroborreliosis that was initially misdiagnosed as post-COVID-19 syndrome. Clinically, the patient presented with a range of chronic and subacute neuropsychiatric symptoms and recalled a tick bite weeks prior to admission. The patient had undergone psychiatric and complementary medical treatments for 1 year before admission and was initially diagnosed with post-COVID-19 syndrome. Admission was performed because of acute worsening with fever, confusion, and unsteady gait. Cerebrospinal fluid (CSF) analysis revealed pleocytosis with elevated borrelia Immunoglobulin M (IgM) and Immunoglobulin M (IgG) CSF/blood antibody indices, indicating acute neuroborreliosis. Anti-NMDA receptor antibodies were identified in the CSF via a cell-based assay and were confirmed by an external laboratory. Other paraneoplastic antibodies were ruled out during in-house examination. Cranial Magnetic resonance imaging (MRI) revealed basal meningitis, rhomb- and limbic encephalitis. A subsequent pelvic Computer tomography (CT) scan identified an ovarian teratoma. The patient's clinical condition improved dramatically with antibiotic treatment and plasmapheresis, the teratoma was surgically removed and she was started on rituximab. Our case highlights that amidst the prevailing focus on COVID-19-related health concerns, other well-established, but rare neurological conditions should not be neglected. Furthermore, our case illustrates that patients may suffer from multiple, concurrent, yet pathophysiologically unrelated neuroinflammatory conditions.

Overcoming data scarcity in biomedical imaging with a foundational multi-task model.

Foundational models, pretrained on a large scale, have demonstrated substantial success across non-medical domains. However, training these models typically requires large, comprehensive datasets, which contrasts with the smaller and more specialized datasets common in biomedical imaging. Here we propose a multi-task learning strategy that decouples the number of training tasks from memory requirements. We trained a universal biomedical pretrained model (UMedPT) on a multi-task database including tomographic, microscopic and X-ray images, with various labeling strategies such as classification, segmentation and object detection. The UMedPT foundational model outperformed ImageNet pretraining and previous state-of-the-art models. For classification tasks related to the pretraining database, it maintained its performance with only 1% of the original training data and without fine-tuning. For out-of-domain tasks it required only 50% of the original training data. In an external independent validation, imaging features extracted using UMedPT proved to set a new standard for cross-center transferability.

Prevalence of Overweight and Obesity, Its Complications, and Progression in a 10-Year Follow-Up in the Gutenberg Health Study (GHS).

INTRODUCTION: Overweight and obesity lead to numerous complications and their treatment. The associated costs represent a health and sociopolitical burden. Therefore, the development of overweight and obesity is of great importance for health policy. METHODS: The Gutenberg Health Study (GHS), a population-based observational study of individuals aged 35-74 years in the city of Mainz and the district of Mainz-Bingen, examined current data on the prevalence and development of overweight and obesity and their association with concomitant diseases and medication use. RESULTS: Among men, 48.1% were overweight and 26.3% had obesity. Among women, these proportions were 32.1% and 24.1%, respectively. Elevated body mass index (BMI) was associated with numerous complications, particularly insulin resistance and type 2 diabetes, arterial hypertension, elevated triglycerides and low HDL cholesterol, and cardiovascular disease. Accordingly, medications to treat these conditions were used significantly more often in individuals with elevated BMI. During the 10-year observation period, mean weight increased in the population. Both men and women had a moderate but significant increase in BMI compared to men and women of the same age at baseline. Individual weight changes over the 10-year observation period, on the other hand, were age-dependent. In the two younger age decades, weight gain was observed, while in the oldest age decade, mean body weight decreased. CONCLUSION: These current data confirm that overweight and obesity are associated with relevant complications and that these complications lead to significant use of appropriate medications. The study also suggests that there is a significant trend toward increased prevalence of obesity (BMI ≥30) over the 10-year period.

The ACROBAT 2022 challenge: Automatic registration of breast cancer tissue.

The alignment of tissue between histopathological whole-slide-images (WSI) is crucial for research and clinical applications. Advances in computing, deep learning, and availability of large WSI datasets have revolutionised WSI analysis. Therefore, the current state-of-the-art in WSI registration is unclear. To address this, we conducted the ACROBAT challenge, based on the largest WSI registration dataset to date, including 4,212 WSIs from 1,152 breast cancer patients. The challenge objective was to align WSIs of tissue that was stained with routine diagnostic immunohistochemistry to its H&E-stained counterpart. We compare the performance of eight WSI registration algorithms, including an investigation of the impact of different WSI properties and clinical covariates. We find that conceptually distinct WSI registration methods can lead to highly accurate registration performances and identify covariates that impact performances across methods. These results provide a comparison of the performance of current WSI registration methods and guide researchers in selecting and developing methods.

Deep feature based cross-slide registration.

Registration of multiple sections in a tissue block is an important pre-requisite task before any cross-slide image analysis. Non-rigid registration methods are capable of finding correspondence by locally transforming a moving image. These methods often rely on an initial guess to roughly align an image pair linearly and globally. This is essential to prevent convergence to a non-optimal minimum. We explore a deep feature based registration (DFBR) method which utilises data-driven descriptors to estimate the global transformation. A multi-stage strategy is adopted for improving the quality of registration. A visualisation tool is developed to view registered pairs of WSIs at different magnifications. With the help of this tool, one can apply a transformation on the fly without the need to generate a transformed moving WSI in a pyramidal form. We compare the performance on our dataset of data-driven descriptors with that of hand-crafted descriptors. Our approach can align the images with only small registration errors. The efficacy of our proposed method is evaluated for a subsequent non-rigid registration step. To this end, the first two steps of the ANHIR winner's framework are replaced with DFBR to register image pairs provided by the challenge. The modified framework produce comparable results to those of the challenge winning team.

Comparison of consecutive and restained sections for image registration in histopathology.

Significance: Although the registration of restained sections allows nucleus-level alignment that enables a direct analysis of interacting biomarkers, consecutive sections only allow the transfer of region-level annotations. The latter can be achieved at low computational cost using coarser image resolutions. Purpose: In digital histopathology, virtual multistaining is important for diagnosis and biomarker research. Additionally, it provides accurate ground truth for various deep-learning tasks. Virtual multistaining can be obtained using different stains for consecutive sections or by restaining the same section. Both approaches require image registration to compensate for tissue deformations, but little attention has been devoted to comparing their accuracy. Approach: We compared affine and deformable variational image registration of consecutive and restained sections and analyzed the effect of the image resolution that influences accuracy and required computational resources. The registration was applied to the automatic nonrigid histological image registration (ANHIR) challenge data (230 consecutive slide pairs) and the hyperparameters were determined. Then without changing the parameters, the registration was applied to a newly published hybrid dataset of restained and consecutive sections (HyReCo, 86 slide pairs, 5404 landmarks). Results: We obtain a median landmark error after registration of 6.5 µm (HyReCo) and 24.1 µm (ANHIR) between consecutive sections. Between restained sections, the median registration error is 2.2 and 0.9 µm in the two subsets of the HyReCo dataset. We observe that deformable registration leads to lower landmark errors than affine registration in both cases (p<0.001), though the effect is smaller in restained sections. Conclusion: Deformable registration of consecutive and restained sections is a valuable tool for the joint analysis of different stains.

Unsupervised many-to-many stain translation for histological image augmentation to improve classification accuracy.

Background: Deep learning tasks, which require large numbers of images, are widely applied in digital pathology. This poses challenges especially for supervised tasks since manual image annotation is an expensive and laborious process. This situation deteriorates even more in the case of a large variability of images. Coping with this problem requires methods such as image augmentation and synthetic image generation. In this regard, unsupervised stain translation via GANs has gained much attention recently, but a separate network must be trained for each pair of source and target domains. This work enables unsupervised many-to-many translation of histopathological stains with a single network while seeking to maintain the shape and structure of the tissues. Methods: StarGAN-v2 is adapted for unsupervised many-to-many stain translation of histopathology images of breast tissues. An edge detector is incorporated to motivate the network to maintain the shape and structure of the tissues and to have an edge-preserving translation. Additionally, a subjective test is conducted on medical and technical experts in the field of digital pathology to evaluate the quality of generated images and to verify that they are indistinguishable from real images. As a proof of concept, breast cancer classifiers are trained with and without the generated images to quantify the effect of image augmentation using the synthetized images on classification accuracy. Results: The results show that adding an edge detector helps to improve the quality of translated images and to preserve the general structure of tissues. Quality control and subjective tests on our medical and technical experts show that the real and artificial images cannot be distinguished, thereby confirming that the synthetic images are technically plausible. Moreover, this research shows that, by augmenting the training dataset with the outputs of the proposed stain translation method, the accuracy of breast cancer classifier with ResNet-50 and VGG-16 improves by 8.0% and 9.3%, respectively. Conclusions: This research indicates that a translation from an arbitrary source stain to other stains can be performed effectively within the proposed framework. The generated images are realistic and could be employed to train deep neural networks to improve their performance and cope with the problem of insufficient numbers of annotated images.

Intermittent Fasting-Short- and Long-Term Quality of Life, Fatigue, and Safety in Healthy Volunteers: A Prospective, Clinical Trial.

Background: Intermittent fasting (IF) is defined as an eating pattern without calorie restrictions, alternating between periods of fasting and eating. In the past decades IF has not only become a popular weight-reducing diet but is thought to improve Quality of Life (QoL) and fatigue. However, very little evidence exists for the general population. Thus, we aimed to assess the impact of a 16-h fasting period per day over a three-month study period on QoL and especially fatigue in healthy people. Methods: We conducted a prospective cohort study including healthy subjects. All participants fasted 16 h for at least five days a week while maintaining their normal lifestyle. In the study, we analysed blood samples as well as QoL through standardized questionnaires (WHO-5 questionnaire, Short Form Health 36). Furthermore, we measured the degree of fatigue with the Fatigue Assessment Scale (FAS) and Fatigue Severity Scale (FSS) as well as compliance, activity records, and weight alterations. All endpoints were evaluated at baseline, after two weeks, four weeks, and three months of IF. Results: A total of 30 participants fasted for the entire study period. The results of the WHO-5 questionnaire (15.6 ± 4.6 vs. 18 ± 3.6, p < 0.0019) demonstrated a significant increase in QoL. For long-term QoL six out of eight domains measured by the Short Form Health 36 (SF-36) significantly improved (e.g., physical health: 92.3 ± 11.6 vs. 96.5 ± 6.3, p = 0.015; mental health: 75.5 ± 12.0 vs. 81.7 ± 9.0; p < 0.001 and body pain: 74.1 ± 31.8 vs. 89.5 ± 14.9; p = 0.008) after three months. Fatigue significantly decreased from 10.3 ± 3.2 to 8.4 ± 2.5; p = 0.002 for mental fatigue and from 12.6 ± 3.8 to 10.7 ± 3.3; p = 0.002 measured by the FAS. The mean FSS-Score at baseline was 3.5 ± 1.2 compared to 2.9 ± 1.1 (scale 1−7) after three months (p < 0.001). Notably, the proliferation marker IGF-1 was significantly reduced. No clinically significant changes in laboratory parameters were observed that would have endangered a participant's safety. Conclusions: IF according to the 16:8 regime over a fasting period of three months significantly improved several aspects of the QoL and decreased fatigue in healthy people, while maintaining a good safety profile. The practicability of this diet was also demonstrated for shift workers and people with a high percentage of active labour. Apart from the improvement in QoL and fatigue, the significant reduction in IGF-1, which can act as an accelerator of tumour development and progression, might be an indicator of the potential benefits of IF for patients with cancer.