2015
DOI: 10.1007/978-3-319-14148-0_11
|View full text |Cite
|
Sign up to set email alerts
|

Automated Radiological Grading of Spinal MRI

Abstract: This paper describes a fully automatic system for obtaining the standard Pfirrmann degeneration grading of individual intervertebral spinal discs in T2 MRI scans. It involves detecting and labeling all the vertebrae in the scan and then learning a regression from the disc region to the grading. In developing the regression function we investigate a spectrum of support regions which involve differing degrees of segmentation of the scan: our intention is to ascertain to what extent segmentation is necessary or d… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
38
0

Year Published

2016
2016
2020
2020

Publication Types

Select...
5
4

Relationship

3
6

Authors

Journals

citations
Cited by 27 publications
(38 citation statements)
references
References 16 publications
0
38
0
Order By: Relevance
“…In comparison, most research on spinal diagnosis classification has been primarily conducted with handcrafted features or what is now referred to as "shallow" learning e.g. in works concerning radiological scoring of the intervertebral discs [3,5,7,9]. One recent successful example of using CNNs on medical images is a segmentation framework proposed by Ronneberger et al [10] which overcame the problem of small amount of data through the use of elastic augmentation though requiring strong (pixel-level) supervision.…”
Section: Why Qualitative Localization?mentioning
confidence: 99%
“…In comparison, most research on spinal diagnosis classification has been primarily conducted with handcrafted features or what is now referred to as "shallow" learning e.g. in works concerning radiological scoring of the intervertebral discs [3,5,7,9]. One recent successful example of using CNNs on medical images is a segmentation framework proposed by Ronneberger et al [10] which overcame the problem of small amount of data through the use of elastic augmentation though requiring strong (pixel-level) supervision.…”
Section: Why Qualitative Localization?mentioning
confidence: 99%
“…This means that if the real degeneration grade was 3, it would never choose 1 or 5 or that it would obtain 100% of sensitivity and specificity with ±1 accuracy. This evaluation strategy was also employed in [17], in which the sensitivity was reported as 85.8% with ±1 accuracy.…”
Section: Discussionmentioning
confidence: 99%
“…In contrast, Lootus et al [17], proposed a method to perform a three level classification of the IVD by using a ±1 precision. Thus, neighbour grades are considered to be the same, reducing the clinical applicability.…”
Section: Grade Imentioning
confidence: 99%
“…Previous works on vertebra segmentation have used a multistep approach incorporating localization of spine and vertebra detection, followed by segmentation [2,3]. Multiple techniques have been applied to vertebra segmentation, such as active shape model and snake-based methods [4,5], level sets, or graph-based techniques using normalized cuts [6,7]. Most of these models depend on prior knowledge in the form of a statistical shape model or spine atlases.…”
Section: Introductionmentioning
confidence: 99%