Cognitive impairment in multiple sclerosis: diagnosis and monitoring.

INTRODUCTION: Cognitive impairment (CI) has a prevalence of 45-70% in people with multiple sclerosis (MS), producing a negative impact on their quality of life, personal life, and work. Early detection of CI has become an important aspect to be considered for an adequate follow-up, to optimize social adaptation and to implement specific cognitive rehabilitation strategies. The aim of this work is to propose a suitable cognitive evaluation of patients with MS based on available and efficient tools for diagnosis and monitoring purposes well supported by literature review and clinical experience. METHODS: A multidisciplinary panel of professionals from the field of neurology, neuropsychology, and neuroimaging performed a literature review of the topic of cognitive impairment assessment. This was combined and completed with their clinical experience to produce a set of recommendations. RESULTS: Some limitations to cognitive evaluation are described: shortage of time and resources during the neurology consultation, scarceness or absence of specialized professionals' availability, importance of tests adaptation, and doubts about its use to define therapeutic efficiency. We recommend a baseline and annual screening evaluation, and we suggest a baseline and periodic neuropsychological assessment. The latter ought to change to a recommendation with the presence of either positive screening test, or subjective to cognitive complaints, screening-test results and patient or family report mismatch, or in specific social/work situations. CONCLUSIONS: Cognitive evaluation should be performed on all patients diagnosed with MS and throughout follow-up. It is necessary to support the creation of multidisciplinary MS teams to optimize the evaluation and follow-up of MS patients.

Using automatically extracted information from mammography reports for decision-support.

OBJECTIVE: To evaluate a system we developed that connects natural language processing (NLP) for information extraction from narrative text mammography reports with a Bayesian network for decision-support about breast cancer diagnosis. The ultimate goal of this system is to provide decision support as part of the workflow of producing the radiology report. MATERIALS AND METHODS: We built a system that uses an NLP information extraction system (which extract BI-RADS descriptors and clinical information from mammography reports) to provide the necessary inputs to a Bayesian network (BN) decision support system (DSS) that estimates lesion malignancy from BI-RADS descriptors. We used this integrated system to predict diagnosis of breast cancer from radiology text reports and evaluated it with a reference standard of 300 mammography reports. We collected two different outputs from the DSS: (1) the probability of malignancy and (2) the BI-RADS final assessment category. Since NLP may produce imperfect inputs to the DSS, we compared the difference between using perfect ("reference standard") structured inputs to the DSS ("RS-DSS") vs NLP-derived inputs ("NLP-DSS") on the output of the DSS using the concordance correlation coefficient. We measured the classification accuracy of the BI-RADS final assessment category when using NLP-DSS, compared with the ground truth category established by the radiologist. RESULTS: The NLP-DSS and RS-DSS had closely matched probabilities, with a mean paired difference of 0.004±0.025. The concordance correlation of these paired measures was 0.95. The accuracy of the NLP-DSS to predict the correct BI-RADS final assessment category was 97.58%. CONCLUSION: The accuracy of the information extracted from mammography reports using the NLP system was sufficient to provide accurate DSS results. We believe our system could ultimately reduce the variation in practice in mammography related to assessment of malignant lesions and improve management decisions.

Prescribed Drugs and Self-Directed Violence: A Descriptive Study in the Spanish Pharmacovigilance Database.

Self-inflicted violence is a major and growing public health problem and its prediction and prevention is challenging for healthcare systems worldwide. Our aim was to identify prescribed drugs associated with self-directed violent behaviors in Spain. A descriptive, longitudinal and retrospective study of spontaneous reports of adverse drug reactions corresponding to self-directed violence was recorded in the Spanish Pharmacovigilance Database (FEDRA®) from 1984 to 31 March 2021. A total of 710 cases were reported in the study period. The mean age was 45.52 years (range 1-94). There were no gender differences except in children, where most reports were of male children. The main therapeutic groups that were involved included drugs for the nervous system (64.5%) and anti-infectives for systemic use (13.2%). The most commonly reported drugs were varenicline, fluoxetine, lorazepam, escitalopram, venlafaxine, veralipride, pregabalin, roflumilast and bupropion. There were reports of montelukast, hydroxychloroquine, isotretinoin, methylphenidate, infliximab, natalizumab, ribavirin and efavirenz, which were less known to be involved in self-directed violence. This study shows that self-directed violence is a rare adverse drug reaction, and can be related to the use of some medicines. It is important for healthcare professionals to consider this risk in their clinical praxis, implementing person-centred approaches. Further studies are needed, considering comorbidities and potential interactions.

Reports of Symptoms Associated with Supraventricular Arrhythmias as a Serious Adverse Drug Reaction in the Spanish Pharmacovigilance Database.

This study aimed to determine the type of drugs reported as suspected of causing severe supraventricular arrhythmias from the Spanish Human Pharmacovigilance System database. A total of 1053 reports were analysed, of which 526 (50%) were on men and 516 (49%) were on women. The most affected age group was the over-65s, with 593 reports (56%). Of the 1613 drugs, those belonging to the cardiovascular system (ATC Group C) were the most numerous (414 reports, 26%), with digoxin being the most frequent drug (49 reports, 12%). Other common groups were antiinfectives for systemic use (ATC Group J; 306 reports, 19%), antineoplastic and immunomodulating agents (ATC Group L; 198 reports, 12%), and nervous system drugs (ATC Group N; 185 reports, 11%). The most common supraventricular arrhythmia was atrial fibrillation (561 reports, 51%). Regarding outcomes, 730 (66%) patients recovered, 76 (7%) did not recover, 25 (3%) recovered but with sequelae, and 23 (2%) resulted in death. This study revealed that certain drugs have reported to be associated more frequently to supraventricular arrhythmias as serious adverse reactions, especially in the older population. Proper clinical management and effective strategies to ensure medication appropriateness should always be considered to improve patient safety when prescribing drugs.

Treatment of multiple sclerosis with rituximab: A Spanish multicenter experience.

Introduction: Rituximab (RTX) is considered a potential therapeutic option for relapsing-remitting (RRMS) and progressive forms (PMS) of multiple sclerosis (MS). The main objective of this work was to investigate the effectiveness and safety of rituximab in MS. Patients and methods: Observational multicenter study of clinical and radiological effectiveness and safety of rituximab in RRMS and PMS. Results: A total of 479 rituximab-treated patients were included in 12 Spanish centers, 188 RRMS (39.3%) and 291 (60.7%) PMS. Despite standard treatment, the annualized relapse rate (ARR) the year before RTX was 0.63 (SD: 0.8) and 156 patients (41%) had at least one gadolinium-enhanced lesion (GEL) on baseline MRI. Mean EDSS had increased from 4.3 (SD: 1.9) to 4.8 (SD: 1.7) and almost half of the patients (41%) had worsened at least one point. After a median follow-up of 14.2 months (IQR: 6.5-27.2), ARR decreased by 85.7% (p < 0.001) and GEL by 82.9%, from 0.41 to 0.07 (p < 0.001). A significant decrease in EDSS to 4.7 (p = 0.046) was observed after 1 year of treatment and this variable remained stable during the second year of therapy. There was no evidence of disease activity in 68% of patients. Infusion-related symptoms were the most frequent side effect (19.6%) and most were mild. Relevant infections were reported only in 18 patients (including one case of probable progressive multifocal leukoencephalopathy). Conclusion: Rituximab could be an effective and safe treatment in RRMS, including aggressive forms of the disease. Some selected PMS patients could also benefit from this treatment.

Effectiveness and Safety of Teriflunomide in Relapsing-Remitting Multiple Sclerosis and Improvements in Quality of Life: Results from the Real-World TERICARE Study.

INTRODUCTION: Teriflunomide is a once-daily oral immunomodulator approved for relapsing forms of multiple sclerosis (MS) or relapsing-remitting multiple sclerosis (RRMS; depending on the local label), based on extensive evidence from clinical trials and a real-world setting on efficacy, tolerability and patient-reported benefits. The TERICARE study assessed the impact of teriflunomide treatment over 2 years on health-related quality of life (HRQoL) and some of the most common and disabling symptoms of MS, such as fatigue and depression. METHODS: This prospective observational study in Spain included RRMS patients treated with teriflunomide for ≤ 4 weeks. The following patient-reported outcomes (PROs) were collected at baseline and every 6 months for 2 years: the 29-item Multiple Sclerosis Impact Scale version 2 (MSIS-29), the 21-item Modified Fatigue Impact Scale (MFIS-21), the Beck Depression Inventory (BDI-II), the Short Form (SF)-Qualiveen and the Treatment Satisfaction Questionnaire for Medication v1.4 (TSQM). Annualised relapse rate (ARR), disability progression according to the Expanded Disability Status Scale (EDSS), and no evidence of disease activity (NEDA-3) were also assessed. RESULTS: A total of 325 patients were analysed. Patients had a mean (SD) age of 43.2 years (10.4), a mean baseline EDSS score of 1.75 (1.5), a mean number of relapses in the past 2 years of 1.5 (0.7), and 64% had received prior disease-modifying therapy (DMT). Patients showed significant improvements in the psychological domain of MSIS-29 from 35.9 (26.6) at baseline to 29.4 (25.5) at 18 months (p = 0.004) and 29.0 (24.6) at 24 months (p = 0.002). Levels of fatigue and depression were also reduced. After 2 years of treatment with teriflunomide, ARR was reduced to 0.17 (95% CI 0.14-0.21) from the baseline of 0.42 (95% CI 0.38-0.48), representing a 60.1% reduction. Mean EDSS scores remained stable during the study, and 79.9% of patients showed no disability progression. 54.7% of patients achieved NEDA-3 in the first 12 months, which increased to 61.4% during months 12-24. Patients reported increased satisfaction with treatment over the course of the study, regardless of whether they were DMT naive or not. CONCLUSION: Teriflunomide improves psychological aspects of HRQoL and maintains low levels of fatigue and depression. Treatment with teriflunomide over 2 years is effective in reducing ARR and disability progression.

Case Report: Exacerbation of Relapses Following mRNA COVID-19 Vaccination in Multiple Sclerosis: A Case Series.

Introduction: mRNA coronavirus disease 2019 (COVID-19) vaccination has been widely used to arrest the spread of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic. Rarely, autoimmune events such as relapses in patients with multiple sclerosis (MS) have been reported after vaccination. However, the possible effects of vaccination in a patient already experiencing the symptoms of a relapse represent an unusual scenario that has not been described. Patients and Methods: This is a retrospective case series of four patients from three major tertiary referral centers that received mRNA COVID-19 vaccination after starting with symptoms of acute demyelination of the central nervous system due to non-recognized MS. A detailed description of each case, including MRI studies, serum light-neurofilament levels, and cerebrospinal fluid (CSF) cytokine profile, is provided. Case Description: All patients presented exacerbation of ongoing symptoms after vaccination (range 14-112 days first dose). All patients presented MRI features suggestive of highly active MS and fulfilled McDonald 2017 criteria at the time of presentation. All patients presented high serum light-neurofilament levels and oligoclonal G bands restricted to the CSF. Higher levels of interleukin-6 in the CSF were present in the more severe cases. Discussion: We describe exacerbation of relapses after mRNA COVID-19 vaccination. We hypothesize RNA sensors such as Toll-like receptor 7 may be activated and contribute to amplify the inflammatory response during a relapse. Conclusion: Patients should seek medical attention if experiencing acute neurological symptoms, especially before vaccination. Fast diagnostic procedures and prompt treatment should be performed in these patients. Pharmacovigilance and further study are warranted to confirm causality.

Baseline Inflammatory Status Reveals Dichotomic Immune Mechanisms Involved In Primary-Progressive Multiple Sclerosis Pathology.

Objective: To ascertain the role of inflammation in the response to ocrelizumab in primary-progressive multiple sclerosis (PPMS). Methods: Multicenter prospective study including 69 patients with PPMS who initiated ocrelizumab treatment, classified according to baseline presence [Gd+, n=16] or absence [Gd-, n=53] of gadolinium-enhancing lesions in brain MRI. Ten Gd+ (62.5%) and 41 Gd- patients (77.4%) showed non-evidence of disease activity (NEDA) defined as no disability progression or new MRI lesions after 1 year of treatment. Blood immune cell subsets were characterized by flow cytometry, serum immunoglobulins by nephelometry, and serum neurofilament light-chains (sNfL) by SIMOA. Statistical analyses were corrected with the Bonferroni formula. Results: More than 60% of patients reached NEDA after a year of treatment, regardless of their baseline characteristics. In Gd+ patients, it associated with a low repopulation rate of inflammatory B cells accompanied by a reduction of sNfL values 6 months after their first ocrelizumab dose. Patients in Gd- group also had low B cell numbers and sNfL values 6 months after initiating treatment, independent of their treatment response. In these patients, NEDA status was associated with a tolerogenic remodeling of the T and innate immune cell compartments, and with a clear increase of serum IgA levels. Conclusion: Baseline inflammation influences which immunological pathways predominate in patients with PPMS. Inflammatory B cells played a pivotal role in the Gd+ group and inflammatory T and innate immune cells in Gd- patients. B cell depletion can modulate both mechanisms.

Image-guided convection-enhanced delivery of GDNF protein into monkey putamen.

Recently, we developed an MRI-based method that enables tracking of parenchymal infusions of therapeutic agents by inclusion of a contrast reagent in the infusate. We show that both liposomal Gadoteridol (GDL) and free Gadoteridol (Gd) can be used for MRI-monitored infusions into the non-human primate (NHP) putamen to predict the distribution of GDNF protein after convection-enhanced delivery (CED). GDNF and both MRI tracers showed good co-distribution within the putamen and other brain regions. Although the CED infusion technique can distribute GDNF protein over large brain regions, continuous administration of GDNF could cause undesired effects that could counteract the benefits of CED as demonstrated in this study when large volumes of GDNF were delivered that lead to GDNF leakage into CSF. These limitations can be addressed by employing an intermittent CED schedule that permits consistent target coverage without GDNF leakage into CSF or white matter. We present an approach intracranial GDNF infusions that can be optimized by means of real-time monitoring via MRI. Adoption of this new standard, along with advanced, reflux-resistant cannulae, may permit reconsideration of direct GDNF infusion into parenchyma as a clinical strategy, since previous clinical studies involving chronic infusion of recombinant glial cell line-derived neurotrophic factor (GDNF) to the putamen for the treatment of Parkinson's disease have yielded mixed results, a state of affairs that may in part be attributed to suboptimal infusion parameters.

Optimal region of the putamen for image-guided convection-enhanced delivery of therapeutics in human and non-human primates.

Optimal results in the direct brain delivery of brain therapeutics such as growth factors or viral vector into primate brain depend on reproducible distribution throughout the target region. In the present study, we retrospectively analyzed MRI of 25 convection-enhanced delivery (CED) infusions with MRI contrast into the putamen of non-human primates (NHP). Infused volume (V(i)) was compared to total volume of distribution (V(d)) versus V(d) within the target putamen. Excellent distribution of contrast agent within the putamen was obtained in eight cases that were used to define an optimal target volume or "green" zone. Partial or poor distribution with leakage into adjacent anatomical structures was noted in 17 cases, defining "blue" and "red" zones, respectively. Quantitative containment (99±1%) of infused gadoteridol within the putamen was obtained when the cannula was placed in the green zone, 87±3% in the blue zone and 49±0.05% in the red zone. These results were used to determine a set of 3D stereotactic coordinates that define an optimal site for putaminal infusions in NHP and human putamen. We conclude that cannula placement and definition of optimal (green zone) stereotactic coordinates have important implications in ensuring effective delivery of therapeutics into the putamen utilizing routine stereotactic MRI localization procedures and should be considered when local therapies such as gene transfer or protein administration are being translated into clinical therapy.

Time Constraint Increases Mental Load and Influences in the Performance in Small-Sided Games in Basketball.

Purpose: In team sports like basketball, small-sided games have received much scientific attention because of their capacity to simulate competitive environments. The purpose of this study was to explore the impact of different task constraints on mental load, and its consequences for individual and team performance in basketball small-sided games. Method: Twenty-four university students voluntarily took part in this study (Mage = 20.81 years [± 1.76]). Four different 3 × 3 basketball games were played: (A) free game, (B) seven seconds possession, (C) three passes maximum per attack phase, (D) conditions B and C combined. Results: Statistical analysis showed that constraints B, C and D produce more mental load, provoke differences in the affect response and the performance is worse. Conclusion: These results demonstrate the mental cost of restricting the time and number of passes, separately and combined to decide and act.

Brain region volumes and their relationship with disability progression and cognitive function in primary progressive multiple sclerosis.

BACKGROUND AND PURPOSE: Evidence on regional changes resulting from neurodegenerative processes underlying primary progressive multiple sclerosis (PPMS) is still limited. We assessed brain region volumes and their relationship with disability progression and cognitive function in PPMS patients. METHODS: This was an MRI analysis of 43 patients from the prospective Understanding Primary Progressive Multiple Sclerosis (UPPMS) cohort study. MRI scans were performed within 3 months before enrollment and at month 12. RESULTS: Gray matter volume of declive and white matter volumes adjacent to left straight gyrus, right calcarine sulcus, and right inferior occipital gyrus significantly decreased from baseline to month 12. Baseline white matter volumes adjacent to right amygdala and left cuneus significantly differed between patients with and without disability progression, as well as baseline gray matter volumes of left cuneus, right parahippocampal gyrus, right insula, left superior frontal gyrus, declive, right inferior temporal gyrus, right superior temporal gyrus (pole), and right calcarine sulcus. Baseline gray matter volumes of right cuneus and right superior temporal gyrus positively correlated with 12-month Selective Reminding Test and Word List Generation performance, respectively. Gray matter changes in right superior semilunar lobe and white matter adjacent to left declive and right cerebellar tonsil also positively correlated with Word List Generation scores, while white matter change in left inferior semilunar lobe positively correlated with Symbol Digit Modalities Test performance after 12 months. CONCLUSIONS: White and gray matter volumes of specific brain regions could predict disability progression and cognitive performance of PPMS patients after one year.

Short-term data on disease activity, cognition, mood, stigma and employment outcomes in a cohort of patients with primary progressive multiple sclerosis (UPPMS study).

BACKGROUND: Primary progressive multiple sclerosis (PPMS) has long been defined by progressive disability accrual in the absence of initial relapses. However, its underlying neurodegenerative process seems to be accompanied by central nervous system inflammation. A new classification defined multiple sclerosis courses according to clinical/radiological activity and progression. We provide further insight into PPMS activity according to this classification and other daily living aspects. METHODS: This was a multicentre, prospective, cohort study including 55 adult patients with PPMS according to 2010 McDonald criteria, within ten years from neurologic symptom onset and not receiving disease-modifying therapies during the past six months, who were followed up for 12 months. The primary study endpoint was the percentage of patients with active disease based on clinical relapses and/or magnetic resonance activity. Disability progression, cognitive function, physical/psychological impact, depression symptoms, stigma and employment were secondary endpoints. RESULTS: Eleven (25.6%) patients exhibited multiple sclerosis activity throughout the 12-month study follow-up. Fourteen showed non-active multiple sclerosis without progression, 11 non-active multiple sclerosis with progression, 6 active multiple sclerosis without progression and 4 active multiple sclerosis with progression; one patient with disease activity was not assessable for progression. Cognitive function scores remained unchanged or increased, disease physical impact was maintained and disease psychological impact significantly decreased. The proportion of patients with depression symptoms or stigma remained without significant changes as well as employment outcomes. CONCLUSION: This study shows that one-fourth of PPMS patients may exhibit disease activity over one year, with disability progression in approximately one-third but without worsening of cognitive function, disease impact, depression, stigma or employment outcomes.

Effect of Ocrelizumab in Blood Leukocytes of Patients With Primary Progressive MS.

OBJECTIVE: To analyze the changes induced by ocrelizumab in blood immune cells of patients with primary progressive MS (PPMS). METHODS: In this multicenter prospective study including 53 patients with PPMS who initiated ocrelizumab treatment, we determined effector, memory, and regulatory cells by flow cytometry at baseline and after 6 months of therapy. Wilcoxon matched paired tests were used to assess differences between baseline and 6 months' results. p Values were corrected using the Bonferroni test. RESULTS: Ocrelizumab reduced the numbers of naive and memory B cells (p < 0.0001) and those of B cells producing interleukin (IL)-6, IL-10, granulocyte-macrophage colony-stimulating factor (GM-CSF), and tumor necrosis factor-alpha (TNFα) (p < 0.0001 in all cases). By contrast, the proportions of plasmablasts and B cells producing GM-CSF and TNFα increased significantly, suggesting the need for treatment continuation. We also observed a decrease in CD20+ T-cell numbers (p < 0.0001) and percentages (p < 0.0001), and a clear remodeling of the T-cell compartment characterized by relative increases of the naive/effector ratios in CD4+ (p = 0.002) and CD8+ (p = 0.002) T cells and relative decreases of CD4+ (p = 0.03) and CD8+ (p = 0.004) T cells producing interferon-gamma. Total monocyte numbers increased (p = 0.002), but no changes were observed in those producing inflammatory cytokines. The immunologic variations were associated with a reduction of serum neurofilament light chain (sNfL) levels (p = 0.008). The reduction was observed in patients with Gd-enhanced lesions at baseline and in Gd- patients with baseline sNfL >10 pg/mL. CONCLUSIONS: In PPMS, effector B-cell depletion changed T-cell response toward a low inflammatory profile, resulting in decreased sNfL levels.

Natural Language Generation Model for Mammography Reports Simulation.

Extending the size of labeled corpora of medical reports is a major step towards a successful training of machine learning algorithms. Simulating new text reports is a key solution for reports augmentation, which extends the cohort size. However, text generation in the medical domain is challenging because it needs to preserve both content and style that are typical for real reports, without risking the patients' privacy. In this paper, we present a conditioned LSTM-RNN architecture for simulating realistic mammography reports. We evaluated the performance by analyzing the characteristics of the simulated reports and classifying them into benign and malignant classes. An average classification AUC was calculated over two distinct test sets. A qualitative analysis was also performed in which a masked radiologist classified 0.75 of the simulated reports as real reports, showing that both the style and content of the simulated reports were similar to real reports. Finally, we compared our RNN-LSTM generative model with Markov Random Fields. The RNN-LSTM provided significantly better and more stable performance than MRFs ( , Wilcoxon).

CSF chitinase 3-like-1 association with disability of primary progressive MS.

OBJECTIVE: To assess the role of CSF chitinase 3-like-1 (CHI3L1), chitinase 3-like-2 (CHI3L2), and neurofilament light chain (NfL) in predicting the course of primary progressive MS (PPMS). METHODS: We analyzed CSF CHI3L1, CHI3L2, and NfL levels in 25 patients with PPMS with disease duration ≤10 years and no disease-modifying therapy for ≥6 months from the prospective Understanding Primary Progressive Multiple Sclerosis cohort study. CSF samples taken at disease diagnosis were analyzed using commercial ELISAs and following the manufacturer's instructions. Data on Expanded Disability Status Scale (EDSS) scores, disability progression, and cognitive function according to the Brief Repeatable Neuropsychological Battery were also assessed throughout the 1-year study follow-up. RESULTS: Increasing CHI3L1 levels correlated with higher EDSS scores at baseline (ρ = 0.490, 95% CI 0.118-0.742, p = 0.013) and month 12 (ρ = 0.455, 95% CI 0.063-0.725, p = 0.026) and tended to be associated with a higher risk of disability progression according to EDSS scores (OR = 1.008, 95% CI 0.999-1.017, p = 0.089). Increasing CHI3L2 levels also tended to correlate with lower baseline EDSS scores (ρ = -0.366, 95% CI -0.676-0.054, p = 0.086). There was no correlation with regard to NfL levels. CONCLUSIONS: This analysis supports the association between CSF CHI3L1 levels and neurologic disability according to EDSS scores in patients with PPMS. Other chitinase-like proteins such as CHI3L2 may also be involved. CLASSIFICATION OF EVIDENCE: This study provides Class II evidence that CSF CHI3L1 is associated with neurologic disability in patients with PPMS.

A Probabilistic Model to Support Radiologists' Classification Decisions in Mammography Practice.

We developed a probabilistic model to support the classification decisions made by radiologists in mammography practice. Using the feature observations and Breast Imaging Reporting and Data System (BI-RADS) classifications from radiologists examining diagnostic and screening mammograms, we modeled their decisions to understand their judgments. Our model could help improve the decisions made by radiologists using their own feature observations and classifications while maintaining their observed sensitivities. Based on 112,433 mammographic cases from 36,111 patients and 13 radiologists at 2 separate institutions with a 1.1% prevalence of malignancy, we trained a probabilistic Bayesian network (BN) to estimate the malignancy probabilities of lesions. For each radiologist, we learned an observed probabilistic threshold within the model. We compared the sensitivity and specificity of each radiologist against the BN model using either their observed threshold or the standard 2% threshold recommended by BI-RADS. We found significant variability among the radiologists' observed thresholds. By applying the observed thresholds, the BN model showed a 0.01% (1 case) increase in false negatives and a 28.9% (3612 cases) reduction in false positives. When using the standard 2% BI-RADS-recommended threshold, there was a 26.7% (47 cases) increase in false negatives and a 47.3% (5911 cases) reduction in false positives. Our results show that we can significantly reduce screening mammography false positives with a minimal increase in false negatives. We find that learning radiologists' observed thresholds provides valuable information regarding the conservativeness of clinical practice and allows us to quantify the variability in sensitivity across and within institutions. Our model could provide support to radiologists to improve their performance and consistency within mammography practice.

Multi-centre validation of a flow cytometry method to identify optimal responders to interferon-beta in multiple sclerosis.

BACKGROUND AND OBJECTIVES: Percentages of blood CD19+CD5+ B cells and CD8+perforin+ T lymphocytes can predict response to Interferon (IFN)-beta treatment in relapsing-remitting multiple sclerosis (RRMS) patients. We aimed to standardize their detection in a multicenter study, prior to their implementation in clinical practice. METHODS: Fourteen hospitals participated in the study. A reference centre was established for comparison studies. Peripheral blood cells of 105 untreated RRMS patients were studied. Every sample was analyzed in duplicate in the participating centre and in the reference one by flow cytometry. When needed, participating centres corrected fluorescence compensations and negative cut-off position following reference centre suggestions. Concordance between results obtained by participating centres and by reference one was evaluated by intraclass correlation coefficients (ICC) and Spearman correlation test. Centre performance was measured by using z-scores values. RESULTS: After results review and corrective actions implementation, overall ICC was 0.86 (CI: 0.81-0.91) for CD19+CD5+ B cell and 0.89 (CI: 0.85-0.93) for CD8+ perforin+ T cell quantification; Spearman r was 0.92 (0.89-0.95; p <0.0001) and 0.92 (0.88-0.95; p <0.0001) respectively. All centres obtained z-scores≤0.5 for both biomarkers. CONCLUSION: Homogenous percentages of CD19+CD5+ B cells and CD8 perforin+ T lymphocytes can be obtained if suitable compensation values and negative cut-off are pre-established.

Differential Data Augmentation Techniques for Medical Imaging Classification Tasks.

Data augmentation is an essential part of training discriminative Convolutional Neural Networks (CNNs). A variety of augmentation strategies, including horizontal flips, random crops, and principal component analysis (PCA), have been proposed and shown to capture important characteristics of natural images. However, while data augmentation has been commonly used for deep learning in medical imaging, little work has been done to determine which augmentation strategies best capture medical image statistics, leading to more discriminative models. This work compares augmentation strategies and shows that the extent to which an augmented training set retains properties of the original medical images determines model performance. Specifically, augmentation strategies such as flips and gaussian filters lead to validation accuracies of 84% and 88%, respectively. On the other hand, a less effective strategy such as adding noise leads to a significantly worse validation accuracy of 66%. Finally, we show that the augmentation affects mass generation.