Automated assessment of new and enlarged white matter and cortical lesions in early multiple sclerosis
Author(s): ,
M.J. Fartaria
Affiliations:
Advanced Clinical Imaging Technology, Siemens Healthcare AG; Department of Radiology, Centre Hospitalier Universitaire Vaudois (CHUV) and University of Lausanne (UNIL); Signal Processing Laboratory (LTS 5), École Polytechnique Fédérale de Lausanne (EPFL), Lausanne
,
G. Bonnier
Affiliations:
Translational Imaging in Neurology (ThINk) Basel, Department of Medicine and Biomedical Engineering, University Hospital Basel and University of Basel, Basel
,
T. Kober
Affiliations:
Advanced Clinical Imaging Technology, Siemens Healthcare AG; Department of Radiology, Centre Hospitalier Universitaire Vaudois (CHUV) and University of Lausanne (UNIL); Signal Processing Laboratory (LTS 5), École Polytechnique Fédérale de Lausanne (EPFL), Lausanne
,
M. Bach Cuadra
Affiliations:
Department of Radiology, Centre Hospitalier Universitaire Vaudois (CHUV) and University of Lausanne (UNIL); Signal Processing Laboratory (LTS 5), École Polytechnique Fédérale de Lausanne (EPFL), Lausanne; Medical Image Analysis Laboratory (MIAL), Centre d`Imagerie BioMédicale (CIBM), Lausanne
C. Granziera
Affiliations:
Translational Imaging in Neurology (ThINk) Basel, Department of Medicine and Biomedical Engineering, University Hospital Basel and University of Basel, Basel; Neurologic Clinic and Policlinic, Departments of Medicine, Biomedicine, and Clinical Research, University Hospital Basel, University of Basel; Department of Biomedical Engineering, University of Basel, Basel, Switzerland
ECTRIMS Online Library. Fartaria M. Oct 12, 2018; 229577; P1033
Mário João Fartaria
Mário João Fartaria
Contributions
×
Abstract

Abstract: P1033

Type: Poster Sessions

Abstract Category: Clinical aspects of MS - Clinical assessment tools

Background: New and enlarged T2-hyperintense lesions located in white matter (WM) and cortex are biomarkers of MS disease progression and therapy response. Yet, the manual detection of those lesions is challenging, time-consuming and suffers from intra- and inter-rater variability.
Objectives: To evaluate the prototype LeMan-PV for automated detection of new and enlarged lesions. LeMan-PV has already shown high sensitivity to WM and cortical lesions in cross-sectional studies and was developed to detect very small lesions because partial-volume effects are taken into account.
Material and methods: MR images were acquired at enrolment (TP1) and at two-years follow-up (TP2) in 32 patients (13 males, 19 females, age range: 20-60 years) with early RRMS (6 years disease duration) on a clinical 3T MRI scanner (MAGNETOM Trio a Tim system, Siemens Healthcare, Erlangen, Germany). Patients had EDSS scores between 1.5 and 2.5, and no evolution in EDSS was measured over 2 years. As a part of the MRI protocol, MP2RAGE, 3D FLAIR and 3D DIR images (voxel size=1.1x1.0x1.2mm3) were obtained. Manual segmentations of new and enlarged MS lesions were performed by one expert neurologist and a radiologist, and used as ground truth (GT). Lesions were classified using the following criteria: i) new - lesion identifiable on TP2 but not in TP1; ii) enlarged - lesion with increased diameter by at least 50% in TP2 with respect to TP1. LeMan-PV was evaluated against the GT for WM and cortical lesions using the following metrics: true positive rate (TPR, number of detected lesions/total GT lesions) and number of false positives.
Results: LeMan-PV achieved a median TPR of 80% (range 33-100%) and 56% (range 0-100%) respectively for WM and cortical new/enlarged lesions. Most of the missed lesions were very small lesions with the size bellow the mean lesion size (≈ 60 µl). Overall, the method showed a median of 0 false positives (range 0-9).
Discussion and conclusion: LeMan-PV is highly specific and exhibits a good performance in the detection of new/enlarged WM lesions. Yet, despite we used MR sequences with high sensitivity to cortical lesions, LeMan-PV shows a lower performance in the detection of new/enlarged cortical lesions due to their low occurrence in our dataset, and their small size and low contrast-to-noise ratio.
* The last two authors contributed equally to this work
Disclosure: M.J. Fartaria, and T. Kober are employees of Siemens Healthineers.

By clicking “Accept Terms & all Cookies” or by continuing to browse, you agree to the storing of third-party cookies on your device to enhance your user experience and agree to the user terms and conditions of this learning management system (LMS).

Cookie Settings
Accept Terms & all Cookies