You need JavaScript to view this

Classification of brain compartments and head injury lesions by neural networks applied to MRI

Abstract

An automatic, neural network-based approach was applied to segment normal brain compartments and lesions on MR images. Two supervised networks, backpropagation (BPN) and counterpropagation, and two unsupervised networks, Kohonen learning vector quantizer and analog adaptive resonance theory, were trained on registered T2-weighted and proton density images. The classes of interest were background, gray matter, white matter, cerebrospinal fluid, macrocystic encephalomalacia, gliosis, and `unknown`. A comprehensive feature vector was chosen to discriminate these classes. The BPN combined with feature conditioning, multiple discriminant analysis followed by Hotelling transform, produced the most accurate and consistent classification results. Classifications of normal brain compartments were generally in agreement with expert interpretation of the images. Macrocystic encephalomalacia and gliosis were recognized and, except around the periphery, classified in agreement with the clinician`s report used to train the neural network. (orig.)
Authors:
Kischell, E R; [1]  Kehtarnavaz, N; [1]  Hillman, G R; [2]  Levin, H; [3]  Lilly, M; [3]  Kent, T A [4] 
  1. Dept. of Electrical Engineering, Texas A and M Univ., College Station, TX (United States)
  2. Dept. of Pharmacology, Univ. of Texas Medical Branch, Galveston, TX (United States)
  3. Dept. of Neurosurgery, Univ. of Texas Medical Branch, Galveston, TX (United States)
  4. Dept. of Neurology and Psychiatry, Univ. of Texas Medical Branch, Galveston, TX (United States)
Publication Date:
Oct 01, 1995
Product Type:
Journal Article
Reference Number:
SCA: 550602; PA: DEN-96:0F8978; EDB-96:089519; SN: 96001592976
Resource Relation:
Journal Name: Neuroradiology; Journal Volume: 37; Journal Issue: 7; Other Information: PBD: Oct 1995
Subject:
55 BIOLOGY AND MEDICINE, BASIC STUDIES; NMR IMAGING; BRAIN; HEAD; INJURIES; IMAGE PROCESSING; NEURAL NETWORKS; PATIENTS; DIAGNOSTIC TECHNIQUES
OSTI ID:
228725
Country of Origin:
Germany
Language:
English
Other Identifying Numbers:
Journal ID: NRDYAB; ISSN 0028-3940; TRN: DE96F8978
Submitting Site:
DEN
Size:
pp. 535-541
Announcement Date:

Citation Formats

Kischell, E R, Kehtarnavaz, N, Hillman, G R, Levin, H, Lilly, M, and Kent, T A. Classification of brain compartments and head injury lesions by neural networks applied to MRI. Germany: N. p., 1995. Web. doi:10.1007/s002340050151.
Kischell, E R, Kehtarnavaz, N, Hillman, G R, Levin, H, Lilly, M, & Kent, T A. Classification of brain compartments and head injury lesions by neural networks applied to MRI. Germany. doi:10.1007/s002340050151.
Kischell, E R, Kehtarnavaz, N, Hillman, G R, Levin, H, Lilly, M, and Kent, T A. 1995. "Classification of brain compartments and head injury lesions by neural networks applied to MRI." Germany. doi:10.1007/s002340050151. https://www.osti.gov/servlets/purl/10.1007/s002340050151.
@misc{etde_228725,
title = {Classification of brain compartments and head injury lesions by neural networks applied to MRI}
author = {Kischell, E R, Kehtarnavaz, N, Hillman, G R, Levin, H, Lilly, M, and Kent, T A}
abstractNote = {An automatic, neural network-based approach was applied to segment normal brain compartments and lesions on MR images. Two supervised networks, backpropagation (BPN) and counterpropagation, and two unsupervised networks, Kohonen learning vector quantizer and analog adaptive resonance theory, were trained on registered T2-weighted and proton density images. The classes of interest were background, gray matter, white matter, cerebrospinal fluid, macrocystic encephalomalacia, gliosis, and `unknown`. A comprehensive feature vector was chosen to discriminate these classes. The BPN combined with feature conditioning, multiple discriminant analysis followed by Hotelling transform, produced the most accurate and consistent classification results. Classifications of normal brain compartments were generally in agreement with expert interpretation of the images. Macrocystic encephalomalacia and gliosis were recognized and, except around the periphery, classified in agreement with the clinician`s report used to train the neural network. (orig.)}
doi = {10.1007/s002340050151}
journal = {Neuroradiology}
issue = {7}
volume = {37}
journal type = {AC}
place = {Germany}
year = {1995}
month = {Oct}
}