Segmentation of magnetic resonance images of the thorax by backpropagation

Middleton, Ian; Damper, R. I.

doi:10.1109/icnn.1995.487753

Cited by 6 publications

(4 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This showed that the method of selecting the training data as described in Section 4.2, using all lung-boundary examples plus an equal number of randomly selected non-boundary examples, led to a very poor classification [10]. In classifying the interior contour of the brain with a similar MLP to that used here, Chiou and Hwang [9] also randomly (and manually) selected their training examples and obtained poor quality results (see their Fig.…”

Section: Results Using Squared-error Cost Functionmentioning

confidence: 88%

“…It is noticeable that precision is consistently poorer than recall; the reason for this will become apparent later. This preliminary work also revealed (results not shown) that the performance of the MLP improved with the size of the input window, and that an input window of dimension at least m = 7 was necessary to distinguish the target boundary from the boundaries of non-target objects such as the great blood vessels [10]. Hence, as previously stated, a (7 × 7) input window has been routinely used-since it provides a good compromise between accuracy and computational expense.…”

Section: Iterative Selection Of Training Examplesmentioning

confidence: 93%

“…Early results for the classification stage, using data from a single subject and a restricted number of slices, were reported by Middleton and Damper [10], and showed good segmentation of the lung boundaries in a given MR image of the torso. Unfortunately, however, generalisation to other (unseen) slices and subjects was very much worse.…”

Section: Introductionmentioning

confidence: 91%

See 2 more Smart Citations

Segmentation of magnetic resonance images using a combination of neural networks and active contour models

Middleton

Damper

2004

Medical Engineering & Physics

View full text Add to dashboard Cite

Segmentation of medical images is very important for clinical research and diagnosis, leading to a requirement for robust automatic methods. This paper reports on the combined use of a neural network (a multilayer perceptron, MLP) and active contour model ('snake') to segment structures in magnetic resonance (MR) images. The perceptron is trained to produce a binary classification of each pixel as either a boundary or a non-boundary point. Subsequently, the resulting binary (edge-point) image forms the external energy function for a snake, used to link the candidate boundary points into a continuous, closed contour. We report here on the segmentation of the lungs from multiple MR slices of the torso; lung-specific constraints have been avoided to keep the technique as general as possible. In initial investigations, the inputs to the MLP were limited to normalised intensity values of the pixels from an (7 × 7) window scanned across the image. The use of spatial coordinates as additional inputs to the MLP is then shown to provide an improvement in segmentation performance as quantified using the effectiveness measure (a weighted product of precision and recall). Training sets were first developed using a lengthy iterative process. Thereafter, a novel cost function based on effectiveness is proposed for training that allows us to achieve dramatic improvements in segmentation performance, as well as faster, non-iterative selection of training examples. The classifications produced using this cost function were sufficiently good that the binary image produced by the MLP could be post-processed using an active contour model to provide an accurate segmentation of the lungs from the multiple slices in almost all cases, including unseen slices and subjects.

show abstract

Section: Results Using Squared-error Cost Functionmentioning

confidence: 88%

Section: Iterative Selection Of Training Examplesmentioning

confidence: 93%

See 1 more Smart Citation

Segmentation of magnetic resonance images using a combination of neural networks and active contour models

Middleton

Damper

2004

Medical Engineering & Physics

View full text Add to dashboard Cite

show abstract

“…However, the initial classification achieved by the neural network in this work is relatively poor and requires a complex model-based deformable model to extract the final boundary. Early results for the classification stage, using data from a single subject and a restricted number of slices, were reported by Middleton et al [48], and have shown the better segmentation of the lung boundaries in a given MR image of the torso. Unfortunately, however, generalization to other slices and subjects was very much poor.…”

Section: Applications To Medical Image Analysis Using Parametric Defomentioning

confidence: 99%

External Force for Deformable Models in Medical Image Segmentation: A Survey

Amarapur¹,

Kulkarni²

2011

SIPIJ

View full text Add to dashboard Cite

Many image segmentation techniques are available in the literature. Among the available techniques, parametric deformable models play an important role in many medical imaging applications. These models have been proved to be effective in segmenting an anatomic structure using constraints derived from the image data along with a prior knowledge about the location, size and shape of these structures. These models also support highly intuitive interaction mechanisms, which helps medical practitioners to bring their expertise to bear on the model-based image interpretation task. In this paper, a review of nineteen different external forces for the parametric deformable model applied to the medical image segmentation is presented. The main purpose of survey is to identify and discuss each category with its principle, mathematical model, advantages, disadvantages and applications to medical image analysis.

show abstract