Department of Biomedical Engineering Posters and Presentations
Temporal Salience Measure for Assessing Image-Based Features
Poster Number
98
Document Type
Poster
Publication Date
3-2016
Abstract
Medical imaging is a useful, minimally invasive medical procedure used in diagnostics. For example, mammograms are considered the gold standard in breast cancer diagnostics. However, like most imaging modalities, in order to obtain a diagnosis, the image must be analyzed by a trained professional, a radiologist. Since human interpretation is required, the process of diagnosis becomes subjective and can experience inaccuracies. This subjective evaluation of the quality of an image is based on whether it could present useful information for human. Thus, to quantitatively and objectively evaluate the quality of an image, in previous work, researchers measured the strength of scale-based contrast features based on human visual system (HVS), which is called the most salient features contained within a medical image, and use it to assess the task-based quality of medical image, like mammograms. Yet, the software implementation of this work had been lost for the most part. So firstly, our goal is to re-create the Perconti software, and validate it by using it with the old eye-track data to test the robust of our program. However, when people are doing the detection task, the duration of eye fixation is variant, which means the salience region will not be fixed all the time. To find out the connection of salience area with time variation, we propose to define temporal salience and compute it for a sequence of images using extensions of the original image-based salience measure. In our presentation, we intend on describing the individual components used in defining the salience measure and their respective purposes. We will discuss the process of how we go from the mammogram image to the final salience measure. The process includes the use of several image analysis techniques, from contrast masking, transformations and application of gabor filters. We will define these components and how they work.
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Open Access
1
Temporal Salience Measure for Assessing Image-Based Features
Medical imaging is a useful, minimally invasive medical procedure used in diagnostics. For example, mammograms are considered the gold standard in breast cancer diagnostics. However, like most imaging modalities, in order to obtain a diagnosis, the image must be analyzed by a trained professional, a radiologist. Since human interpretation is required, the process of diagnosis becomes subjective and can experience inaccuracies. This subjective evaluation of the quality of an image is based on whether it could present useful information for human. Thus, to quantitatively and objectively evaluate the quality of an image, in previous work, researchers measured the strength of scale-based contrast features based on human visual system (HVS), which is called the most salient features contained within a medical image, and use it to assess the task-based quality of medical image, like mammograms. Yet, the software implementation of this work had been lost for the most part. So firstly, our goal is to re-create the Perconti software, and validate it by using it with the old eye-track data to test the robust of our program. However, when people are doing the detection task, the duration of eye fixation is variant, which means the salience region will not be fixed all the time. To find out the connection of salience area with time variation, we propose to define temporal salience and compute it for a sequence of images using extensions of the original image-based salience measure. In our presentation, we intend on describing the individual components used in defining the salience measure and their respective purposes. We will discuss the process of how we go from the mammogram image to the final salience measure. The process includes the use of several image analysis techniques, from contrast masking, transformations and application of gabor filters. We will define these components and how they work.
Comments
Presented at: GW Research Days 2016