Multiprotocol MR Image Segmentation in Multiple Sclerosis: Experience with Over 1,000 Studies
Section snippets
T2-Weighted Imaging
The sensitivity of T2-weighted imaging in the detection of MS has long been recognized (5, 6). MS lesions appear hyperintense on T2-weighted images. The clinical course of MS is mainly defined by the baseline disability of the patient, also called the disease burden. The extent of lesions on T2-weighted images, usually expressed as the total volume of lesions (T2LV), is currently regarded as the MR imaging measure of this disease burden. Dual-echo T2-weighted and proton-density (PD)–weighted
T1-Weighted Imaging with Gadolinium Enhancement
In MS, gadolinium (gadopentetate dimeglumine) enhancement on MR images is characterized by transient blood-brain barrier abnormality and inflammation that represents the acute stage of the evolution of MS (17, 18). On enhanced T1-weighted images, gadolinium enhancement appears as a homogeneous, strongly hyperintense lesion or as a ring-shaped area of hyperintensity at the edge of chronic reactivated lesions. Enhancement is sensitive in demonstrating disease activity and may help distinguish
MT Imaging
MS is now believed to be a diffuse process that extensively involves the WM and is not restricted to the focal regions of disease activity visible as “lesions” on conventional T2-weighted, PD, and gadolinium-enhanced T1-weighted images. This notion came about mainly from MT image analysis (22) but is also supported by postmortem study (23) revealing microscopic disease characterized by edema, cellular infiltration, and demyelination in macroscopically normal-appearing WM. In MT imaging (24),
T1-Weighted Imaging
Another standard MR imaging technique routinely used in MS is T1-weighted imaging (spin echo), which often shows hypointense areas that have been suggested to represent areas of axonal loss and gliosis (28). In recent quantitative studies (29, 30), a significant correlation was reported between increase in disability and increase in the volume of these lesions. If hypointense lesions represent areas of severe demyelination, axonal loss, or gliosis (27, 28), then the higher rate of new lesions
Data Acquisition
The MR imaging protocols we have used are listed in the Table. Every patient recruited in our trial undergoes all acquisitions listed. Our image database currently consists of the following number of patient studies and three-dimensional volume images that have been processed by the methods described in this section: 690 T2-weighted imaging studies; 660 T1-weighted studies, each with and without contrast enhancement; and 670 MT studies. Altogether, the study included 100 patients and over 4,000
Validation
Any segmentation effort consists of (a) a theoretical or algorithmic framework, (b) considerable engineering effort to make the framework work in the application at hand, and (c) evaluation to establish the precision, accuracy, and efficiency of the method for the particular application. Precision here refers to the reproducibility of the segmentation results with all subjective actions taken into consideration, including how the patient is positioned in the imager and any operator input
Automation, Failure, and User Assistance
Any segmentation method can go wrong, and therefore will, if enough studies are analyzed in a routine clinical trial. Therefore, for quality assurance, it is important to have a knowledgeable human operator within the processing loop. Complete automation may therefore be an elusive goal, perhaps reachable after only a great deal of experience within the same imaging modality, protocol, and application setting. As researchers and developers, our aim should be to consider human interaction within
Acknowledgments
The authors are grateful to Mary A. Blue for typing the manuscript.
References (54)
- et al.
A histological, histochemical and biochemical study of the macroscopically normal white matter in multiple sclerosis
J Neurol Sci
(1979) - et al.
Fuzzy connectedness and object definition: theory, algorithms, and applications in image segmentation
Graph Models Image Proc
(1996) - et al.
Scale-based fuzzy connected image segmentation: theory, algorithms, and validation
Comput Vis Image Understanding
(2000) Lectures on the diseases of the nervous system
(1977)Geographic distribution of multiple sclerosis: an update with special reference to Europe and the Mediterranean region
Acta Neurol Scand
(1980)Multiple sclerosis and other demyelinating disease
Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS)
Neurology
(1983)- et al.
Nuclear magnetic resonance imaging in multiple sclerosis
Ann Neurol
(1983) - et al.
MR imaging of multiple sclerosis: comparison with clinical and CT examinations in 74 patients
AJR Am J Roentgenol
(1985) - et al.
Intramuscular interferon beta-1a for disease progression in relapsing multiple sclerosis. The Multiple Sclerosis Collaborative Research Group (MSCRG)
Ann Neurol
(1996)
Impact of interferon beta-1a on neurologic disability in relapsing multiple sclerosis. The Multiple Sclerosis Collaborative Research Group (MSCRG)
Neurology
Extended use of glatiramer acetate (Copaxone) is well tolerated and maintains its clinical effect on multiple sclerosis relapse rate and degree of disability. Copolymer 1 Multiple Sclerosis Study Group
Neurology
Neurology
The role of magnetic resonance techniques in understanding and managing multiple sclerosis
Brain
Interferon beta-1b is effective in relapsing-remitting multiple sclerosis. II. MRI analysis results of a multicenter, randomized, double-blind, placebo-controlled trial. UBC MS/MRI Study Group and the IFNB Multiple Sclerosis Study Group
Neurology
Relapsing-remitting multiple sclerosis: longitudinal analysis of MR images—lack of correlation between changes in T2 lesion volume and clinical findings
Radiology
Patterns of disease activity in multiple sclerosis: clinical and magnetic resonance imaging study
Br Med J
Breakdown of the blood-brain barrier precedes symptoms and other MRI signs of new lesions in multiple sclerosis: pathogenetic and clinical implications
Brain
Quantitative brain MRI lesion load predicts the course of clinically isolated syndromes suggestive of multiple sclerosis
Neurology
Multiple sclerosis: gadolinium enhancement in MR imaging
Radiology
Heterogeneity of blood-brain barrier changes in multiple sclerosis: an MRI study with gadolinium-DTPA enhancement
Neurology
Gadolinium enhancement increases the sensitivity of MRI in detecting disease activity in multiple sclerosis
Brain
Comparison of triple dose versus standard dose gadolinium-DTPA for detection of MRI enhancing lesions in patients with MS
Neurology
The site of the lesion causing deafness in multiple sclerosis
Scand Audiol
Experimental allergic encephalomyelitis and multiple sclerosis: lesion characterization with magnetization transfer imaging
Radiology
Magnetization transfer contrast (MTC) and tissue water proton relaxation in vivo
Magn Reson Med
Cited by (0)
Supported in part by grants NS37172 and NS29029 from the National Institutes of Health.