Enhancement and associative restoration of electronic portal images in radiotherapy

https://doi.org/10.1016/S1386-5056(98)00038-0Get rights and content

Abstract

Electronic portal imaging devices use the high energy treatment beam to project the body interior of the patient during radiation onto a fluorescent screen that is scanned by a camera. Because of the imaging physics, the unprocessed images of very poor quality, but they are the only available information during treatment for observation of the patient’s organs. This paper presents an approach that combines an associative restoration algorithm with a fuzzy image enhancement technique. By fusion of the electronic portal image (EPI) with a pre-treatment captured simulator image (SI) a higher image quality than by conventional techniques is achieved.

Introduction

The electronic portal imaging device has become an important tool for the clinician to verify the shape and the location of the therapy beam with respect to the patient’s anatomy [16]. Normally, a visual comparison of the real patient position related to the beam with the planned treatment field is performed. This treatment field is defined during diagnostics and treatment planning. For this purpose, a treatment simulation takes place, as a result of which a simulator image (SI) is captured. Because of the imaging physics the unprocessed electronic portal image (EPI) is very poor in quality-compared with the SI that is usually an X-ray image. However, EPIs are unlike simulator images dynamic (in-treatment) data and therefore the only available information about the real position of the patient and his organs. The conventional EPI allows only a rough verification of patient position relative to bony structures. State of the art conventional enhancement techniques can be applied to EPIs that give some improvement for further visual analysis after the treatment (off-line). Fig. 1 demonstrates the approach that combines an associative restoration algorithm with a fuzzy image enhancement technique, which improves quality. The advantage of a fuzzy approach compared to conventional image enhancement algorithms is its stability under nearly uncertain conditions on the one hand and the simple adjustment to achieve desired effect in the resulting image on the other hand. Before enhancing, the radiation field is segmented by a rule-based approach. This area defines a region of interest on which the following image processing is applied. Because of the different scanning techniques, the SI and EPI possess usually different scaling factors and may be distorted. Before applying the proposed methodology, the required transformations are therefore determined by feature alignment.

The main idea of the associative restoration is the inclusion of a-priori knowledge given by the SI for the restoration of the EPI to achieve a much better in-treatment image and to allow more reliable feature tracking. This is done in two steps. First a specially structured artificial neural network that we call modified associative memory is trained with the SI from pre-treatment. Possible shape variations are included in the training data. In the second step, the recall by the EPI follows in-treatment. In this way, an image with a much higher quality than from conventional solutions is obtained. Variations in shape and position are considered by this method.

Section snippets

Fuzzy image processing

The object of all enhancement techniques is to transform the original image into a result image that is more suitable for further processing steps and/or human perception. For the visual analysis of the EPI by the physician reliable knowledge about the position of the inner organs of the patient within the treatment field is more important than a correct reproduction of the gray-levels intensity distribution. The method should be robust and a certain enhancement effect should be easily

Feature alignment

Because of the differing scanning techniques used, EPIs and SIs present the same patient features (e.g. bony structures, soft tissues, air) in different ways. The position of the patient and of his organs may differ in both images because of the limited accuracy of patient set-up and because of natural changes in anatomy (radiation may be given days after treatment simulation). Locating corresponding features in both images is an essential requirement of the associative approach, as is the case

Associative restoration

The improvement for noise-degraded images obtained with restoration techniques depends essentially on the a-priori information on the statistical properties of the object and the imaging system [5]. The creation of a mathematical imaging model by systematic means is often impracticable and is closely tied to the concrete data. The theory of artificial neural networks provides a powerful method for image restoration when the parameters are trained in such a way that the neural networks

Example

Fig. 11d shows an EPI of the human head and neck of typical quality and Fig. 11a the corresponding radiograph (simulator image) from treatment simulation. The better image quality of the simulator image is obvious, but it is not available during the treatment fraction. Therefore a better quality of EPI that is available on-line is desired.

The fuzzy approach described in Section 2.4was applied to both images yielding images (e) and (b). The region of interest (approximately the radiation field)

Conclusion

The aim of this work is to process the EPI for a better analysis by the physician than is possible with established pre-processing algorithms. The presented approach combines a sophisticated fuzzy enhancement algorithm with an associative memory that uses a priori knowledge from the SI. The visual expressiveness of the EPI is considerably enhanced and a displacement map is obtained. This is of great benefit for the alignment verification of the patient related to the intended position and it

Acknowledgements

This work was supported by EU grant (Contract No BMH4-CT95-0567) and by LSA grant (FKZ: 002 KE 1996).

References (20)

  • A. De Luca et al.

    A definition of nonprobabilistic entropy in the setting of fuzzy set theory

    Inf. Control

    (1972)
  • L.A. Zadeh

    Fuzzy sets

    Inf. Control

    (1965)
  • H.M. Abbas, M.M. Fahmy, A neural model for adaptive Karhuen Loéve transformation (KLT), Int. Joint Conf. on Neural...
  • S. Banks

    Signal Processing, Image Processing and Pattern Recognition

    (1990)
  • M. Craig et al.

    On the use of fuzzy sets in histogram equalization

    Fuzzy Sets Syst.

    (1992)
  • R.C. Gonzalez et al.

    Digital Image Processing

    (1977)
  • A. Kaufmann
    (1975)
  • T. Kohonen, Self-Organization and Associative Memory, third ed., Springer, Berlin,...
  • R. Krishnapuram, J.M. Keller, Fuzzy set theoretic approach to computer vision: an overview, IEEE Int. Conf. on Fuzzy...
  • K.W. Leszczynski et al.

    The enhancement of radiotherapy verification images by an automated edge detection technique

    Med. Phys.

    (1992)
There are more references available in the full text version of this article.

Cited by (0)

View full text