Assessment of automatic decision-support systems for detecting active T2 lesions in multiple sclerosis patients

Background: Active (new/enlarging) T2 lesion counts are routinely used in the clinical management of multiple sclerosis. Thus, automated tools able to accurately identify active T2 lesions would be of high interest to neuroradiologists for assisting in their clinical activity. Objective: To compare the accuracy in detecting active T2 lesions and of radiologically active patients based on different visual and automated methods. Methods: One hundred multiple sclerosis patients underwent two magnetic resonance imaging examinations within 12 months. Four approaches were assessed for detecting active T2 lesions: (1) conventional neuroradiological reports; (2) prospective visual analyses performed by an expert; (3) automated unsupervised tool; and (4) supervised convolutional neural network. As a gold standard, a reference outcome was created by the consensus of two observers. Results: The automated methods detected a higher number of active T2 lesions, and a higher number of active patients, but a higher number of false-positive active patients than visual methods. The convolutional neural network model was more sensitive in detecting active T2 lesions and active patients than the other automated method. Conclusion: Automated convolutional neural network models show potential as an aid to neuroradiological assessment in clinical practice, although visual supervision of the outcomes is still required.

Download Full-text

Disability progression markers over 6–12 years in interferon-β-treated multiple sclerosis patients

Multiple Sclerosis Journal ◽

10.1177/1352458517698052 ◽

2017 ◽

Vol 24 (3) ◽

pp. 322-330 ◽

Cited By ~ 32

Author(s):

Jordi Río ◽

Àlex Rovira ◽

Mar Tintoré ◽

Susana Otero-Romero ◽

Manuel Comabella ◽

...

Keyword(s):

Multiple Sclerosis ◽

Disease Activity ◽

Clinical Activity ◽

Cox Models ◽

Clinical Cohort ◽

T2 Lesions ◽

Magnetic Resonance Imaging Mri ◽

Multiple Sclerosis Patients ◽

Relapsing Remitting Multiple Sclerosis

Objective: To investigate the association between activity during interferon-beta (IFNβ) therapy and disability outcomes in patients with relapsing–remitting multiple sclerosis (RRMS). Methods: A longitudinal study based on two previously described cohorts of IFNβ-treated RRMS patients was conducted. Patients were classified according to clinical activity after 2 years (clinical cohort) or to clinical and radiological activity after 1 year (magnetic resonance imaging (MRI) cohort). Multivariate Cox models were calculated for early disease activity predicting long-term disability. Results: A total of 516 patients from two different cohorts were included in the analyses. Persistent clinical disease activity during the first 2 years of therapy predicted severe long-term disability (clinical cohort). In the MRI cohort, modified Rio score and no or minimal evidence of disease activity (NEDA/MEDA) did not identify patients with risk of Expanded Disability Status Scale (EDSS) worsening. However, a Rio score ≥ 2 (hazard ratio (HR): 3.3, 95% confidence interval (CI): 1.7–6.4); ≥3 new T2 lesions (HR: 2.9, 95% CI: 1.5–5.6); or ≥2 Gd-enhancing lesions (HR: 2.1, 95% CI: 1.1–4) were able to identify patients with EDSS worsening. Conclusion: Although early activity during IFNβ therapy is associated with poor long-term outcomes, minimal degree of activity does not seem to be predictive of EDSS worsening over 6.7-year mean follow-up.

Download Full-text

A fully automated method of human identification based on dental panoramic radiographs using a convolutional neural network

Dentomaxillofacial Radiology ◽

10.1259/dmfr.20210383 ◽

2021 ◽

Author(s):

Young Hyun Kim ◽

Eun-Gyu Ha ◽

Kug Jin Jeon ◽

Chena Lee ◽

Sang-Sun Han

Keyword(s):

Neural Network ◽

Convolutional Neural Network ◽

High Speed ◽

Large Scale ◽

Oral Surgery ◽

Human Identification ◽

Running Time ◽

Automated Method ◽

Image Characteristics ◽

Proposed Model

Objectives: This study aimed to develop a fully automated human identification method based on a convolutional neural network (CNN) with a large-scale dental panoramic radiograph (DPR) dataset. Methods: In total, 2,760 DPRs from 746 subjects who had 2 to 17 DPRs with various changes in image characteristics due to various dental treatments (tooth extraction, oral surgery, prosthetics, orthodontics, or tooth development) were collected. The test dataset included the latest DPR of each subject (746 images) and the other DPRs (2,014 images) were used for model training. A modified VGG16 model with two fully connected layers was applied for human identification. The proposed model was evaluated with rank-1, –3, and −5 accuracies, running time, and gradient-weighted class activation mapping (Grad-CAM)–applied images. Results: This model had rank-1,–3, and −5 accuracies of 82.84%, 89.14%, and 92.23%, respectively. All rank-1 accuracy values of the proposed model were above 80% regardless of changes in image characteristics. The average running time to train the proposed model was 60.9 sec per epoch, and the prediction time for 746 test DPRs was short (3.2 sec/image). The Grad-CAM technique verified that the model automatically identified humans by focusing on identifiable dental information. Conclusion: The proposed model showed good performance in fully automatic human identification despite differing image characteristics of DPRs acquired from the same patients. Our model is expected to assist in the fast and accurate identification by experts by comparing large amounts of images and proposing identification candidates at high speed.

Download Full-text

Development of Autoimmune Thyroid Disease in Multiple Sclerosis Patients Post-Alemtuzumab Improves Treatment Response

The Journal of Clinical Endocrinology & Metabolism ◽

10.1210/clinem/dgaa453 ◽

2020 ◽

Vol 105 (9) ◽

pp. e3392-e3399 ◽

Cited By ~ 1

Author(s):

Alina Sovetkina ◽

Rans Nadir ◽

Antonio Scalfari ◽

Francesca Tona ◽

Kevin Murphy ◽

...

Keyword(s):

Multiple Sclerosis ◽

Thyroid Disease ◽

Autoimmune Thyroid Disease ◽

Graves Disease ◽

Autoimmune Thyroid ◽

Tertiary Referral Center ◽

T2 Lesions ◽

Relapsing Remitting ◽

Edss Score ◽

Multiple Sclerosis Patients

Abstract Context Alemtuzumab is an anti-CD52 monoclonal antibody used in the treatment of relapsing-remitting multiple sclerosis (MS). Between 20% and 40% of alemtuzumab-treated MS patients develop autoimmune thyroid disease (AITD) as a side effect. Objective The objective of this work is to determine whether MS disease progression following alemtuzumab treatment differs in patients who develop AITD compared to those who do not. Design, Setting, and Patients A retrospective analysis of 126 patients with relapsing-remitting MS receiving alemtuzumab from 2012 to 2017 was conducted at a tertiary referral center. Main Outcome Measures Thyroid status, new relapses, Expanded Disability Status Scale (EDSS) score change, and disability progression following alemtuzumab were evaluated. Results Twenty-six percent (33 out of 126, 25 female, 8 male) of alemtuzumab-treated patients developed AITD, 55% of which was Graves disease. EDSS score following alemtuzumab was reduced in patients who developed AITD compared to those who did not (median [interquartile range]; AITD: –0.25 [–1 to 0.5] vs non-AITD: 0 [1-0]. P = .007]. Multivariable regression analysis confirmed that the development of AITD was independently associated with EDSS score improvement (P = .011). Moreover, AITD patients had higher relapse-free survival following alemtuzumab (P = .023). There was no difference in the number of new focal T2 lesions and contrast-enhancing magnetic resonance imaging lesions developed following alemtuzumab between the 2 groups. Conclusion Graves disease was the most common form of AITD developed by MS patients following alemtuzumab. This study suggests that MS patients who develop AITD may have an improved response to alemtuzumab, as measured by reduced disability and lower relapse rate.

Download Full-text

An Adaptive Semi-automated Integrated System for Multiple Sclerosis Lesion Segmentation in Longitudinal MRI Scans Based on a Convolutional Neural Network

10.1007/978-3-030-89128-2_25 ◽

2021 ◽

pp. 256-265

Author(s):

Andreas Georgiou ◽

Christos P. Loizou ◽

Andria Nicolaou ◽

Marios Pantzaris ◽

Constantinos S. Pattichis

Keyword(s):

Neural Network ◽

Multiple Sclerosis ◽

Convolutional Neural Network ◽

Integrated System ◽

Multiple Sclerosis Lesion ◽

Lesion Segmentation ◽

Longitudinal Mri ◽

Mri Scans

Download Full-text

Reducing return of disease activity in patients with relapsing multiple sclerosis transitioned from natalizumab to teriflunomide: 12-month interim results of teriflunomide therapy

Multiple Sclerosis Journal - Experimental Translational and Clinical ◽

10.1177/2055217318824618 ◽

2019 ◽

Vol 5 (1) ◽

pp. 205521731882461

Author(s):

Stanley L Cohan ◽

Keith Edwards ◽

Lindsay Lucas ◽

Tiffany Gervasi-Follmar ◽

Judy O’Connor ◽

...

Keyword(s):

Multiple Sclerosis ◽

Disease Activity ◽

Brain Magnetic Resonance Imaging ◽

T2 Lesions ◽

Disability Status ◽

The Mean ◽

Multiple Sclerosis Patients ◽

Mean Time ◽

Relapsing Multiple Sclerosis

Background Natalizumab is an effective treatment for relapsing multiple sclerosis. Return of disease activity upon natalizumab discontinuance creates the need for follow-up therapeutic strategies. Objective To assess the efficacy of teriflunomide following natalizumab discontinuance in relapsing multiple sclerosis patients. Methods Clinically stable relapsing multiple sclerosis patients completing 12 or more consecutive months of natalizumab, testing positive for anti-John Cunningham virus antibody, started teriflunomide 14 mg/day, 28 ± 7 days after their final natalizumab infusion. Physical examination, Expanded Disability Status Scale, laboratory assessments, and brain magnetic resonance imaging were performed at screening and multiple follow-up visits. Results Fifty-five patients were enrolled in the study. The proportion of patients relapse-free was 0.94, restricted mean time to first gadolinium-enhancing lesion was 10.9 months and time to 3-month sustained disability worsening was 11.8 months. The mean number of new or enlarging T2 lesions per patient at 12 months was 0.42. Exploratory analyses revealed an annualized relapse rate of 0.08, and a proportion of patients with no evidence of disease activity of 0.68. Forty-seven patients (85.5%) reported adverse events, 95% of which were mild to moderate. Conclusions Teriflunomide therapy initiated without natalizumab washout resulted in a low rate of return of disease activity. Clinicians may consider this a worthwhile strategy when transitioning clinically stable patients off natalizumab to another therapy. ClinicalTrials.gov Identifier: NCT01970410

Download Full-text

Domain Randomization for Detection and Position Estimation of Multiples of a Single Object With Applications to Localizing Bolts on Structures

Volume 1: 39th Computers and Information in Engineering Conference ◽

10.1115/detc2019-97393 ◽

2019 ◽

Cited By ~ 1

Author(s):

Ezra Ameperosa ◽

Pranav A. Bhounsule

Keyword(s):

Neural Network ◽

Convolutional Neural Network ◽

Position Estimation ◽

Learning Approaches ◽

Single Object ◽

Automated Method ◽

The Neural Network ◽

Periodic Maintenance ◽

Camera Position ◽

Periodic Replacement

Abstract Periodic replacement of fasteners such as bolts are an integral part of many structures (e.g., airplanes, cars, ships) and require periodic maintenance that may involve either their tightening or replacement. Current manual practices are time consuming and costly especially due to the large number of bolts. Thus, an automated method that is able to visually detect and localize bolt positions would be highly beneficial. In this paper, we demonstrate the use of deep neural network using domain randomization for detecting and localizing multiple bolts on a workpiece. In contrast to previous deep learning approaches that require training on real images, the use of domain randomization allows for all training to be done in simulation. The key idea here is to create a wide variety of computer generated synthetic images by varying the texture, color, camera position and orientation, distractor objects, and noise, and train the neural network on these images such that the neural network is robust to scene variability and hence provides accurate results when deployed on real images. Using domain randomization, we train two neural networks, a faster regional convolutional neural network for detecting the bolt and predicting a bounding box, and a regression convolutional neural network for estimating the x- and y-position of the bolt relative to the coordinates fixed to the workpiece. Our results indicate that in the best case we are able to detect bolts with 85% accuracy and are able to predict the position of 75% of bolts within 1.27 cm. The novelty of this work is in the use of domain randomization to detect and localize: (1) multiples of a single object, and (2) small sized objects (0.6 cm × 2.5 cm).

Download Full-text

Retrospective unbiased plasma lipidomic of progressive multiple sclerosis patients-identifies lipids discriminating those with faster clinical deterioration

Scientific Reports ◽

10.1038/s41598-020-72654-8 ◽

2020 ◽

Vol 10 (1) ◽

Author(s):

Mario Amatruda ◽

Maria Petracca ◽

Maureen Wentling ◽

Benjamin Inbar ◽

Kamilah Castro ◽

...

Keyword(s):

Multiple Sclerosis ◽

Disease Progression ◽

Progressive Multiple Sclerosis ◽

Primary Progressive Multiple Sclerosis ◽

Disease Course ◽

Primary Progressive ◽

T2 Lesions ◽

Relapsing Remitting ◽

One Year ◽

Multiple Sclerosis Patients

Abstract The disease course of patients with a confirmed diagnosis of primary progressive multiple sclerosis (PPMS) is uncertain. In an attempt to identify potential signaling pathways involved in the evolution of the disease, we conducted an exploratory unbiased lipidomic analysis of plasma from non-diseased controls (n = 8) and patients with primary progressive MS (PPMS, n = 19) and either a rapid (PPMS-P, n = 9) or slow (PPMS-NP, n = 10) disease course based on worsening disability and/or MRI-visible appearance of new T2 lesions over a one-year-assessment. Partial least squares-discriminant analysis of the MS/MSALL lipidomic dataset, identified lipids driving the clustering of the groups. Among these lipids, sphingomyelin-d18:1/14:0 and mono-hexosylceramide-d18:1/20:0 were differentially abundant in the plasma of PPMS patients compared to controls and their levels correlated with MRI signs of disease progression. Lyso-phosphatidic acid-18:2 (LPA-18:2) was the only lipid with significantly lower abundance in PPMS patients with a rapidly deteriorating disease course, and its levels inversely correlated with the severity of the neurological deficit. Decreased levels of LPA-18:2 were detected in patients with more rapid disease progression, regardless of therapy and these findings were validated in an independent cohort of secondary progressive (SPMS) patients, but not in a third cohorts of relapsing–remitting (RRMS) patients. Collectively, our analysis suggests that sphingomyelin-d18:1/14:0, mono-hexosylceramide-d18:1/20:0, and LPA-18:2 may represent important targets for future studies aimed at understanding disease progression in MS.

Download Full-text