Knowledge Based Framework

The CAD scheme detailed in this chapter is based on an adaptive framework. An adaptive framework is capable of modifying itself such that it is more suitable to the environment within which it operates. Within the context of CAD, an adaptable component or a framework, attempts to automatically optimize the lesion detection process for a given mammogram. Broadly speaking, an adaptive characteristic can be built into CAD in three different ways: (1) Using a deterministic component; (2) knowledge-based component; (3) with a knowledge-based framework. Each approach may be used in combination with the others. These approaches are summarized below.

1. Deterministic component: This is the most common strategy for introducing adaptability into a CAD scheme. Typically the component method is fixed but parameters are adaptively determined on a per-image basis. The parameter setting is either performed in a deterministic manner, based directly on an observed feature of the image, e.g. variance of gray scales, or empirically through experimentation. In the past, this approach has been applied to each of the three CAD components, e.g., adaptive contrast enhancement methods to perform an optimal contrast enhancement on an image based on local neighborhoods [1, 2], adaptive segmentation techniques utilizing adaptive clustering [3-5], or thresholding techniques [6-8] to segment an image and adaptive classification methods in the reduction of false-positive regions.

2. Knowledge-based component: An alternative strategy for setting a given component's parameters is to learn the optimal parameter settings for an individual or group of images using machine learning techniques. The mapping between the parameter settings and images is achieved using a global image characteristics.

3. Knowledge-based framework: An extension to the knowledge-based component is to use machine learning principles to learn the utility of a particular component technique for an individual or group of images. Such a knowledge-based framework would be capable of drawing on a variety of different techniques to meet the objectives of each CAD component. The framework would support the definition of an optimal pipeline through the CAD pyramid. To date, no research has been presented into the use of knowledge-based frameworks in medical imaging CAD schemes. We now highlight some past research into knowledge-based components used in CAD schemes to put our proposed model in context.

0 0

Post a comment