RESUMO
The objective of this work is to perform image quality assessment (IQA) of eye fundus images in the context of digital fundoscopy with topological data analysis (TDA) and machine learning methods. Eye health remains inaccessible for a large amount of the global population. Digital tools that automize the eye exam could be used to address this issue. IQA is a fundamental step in digital fundoscopy for clinical applications; it is one of the first steps in the preprocessing stages of computer-aided diagnosis (CAD) systems using eye fundus images. Images from the EyePACS dataset were used, and quality labels from previous works in the literature were selected. Cubical complexes were used to represent the images; the grayscale version was, then, used to calculate a persistent homology on the simplex and represented with persistence diagrams. Then, 30 vectorized topological descriptors were calculated from each image and used as input to a classification algorithm. Six different algorithms were tested for this study (SVM, decision tree, k-NN, random forest, logistic regression (LoGit), MLP). LoGit was selected and used for the classification of all images, given the low computational cost it carries. Performance results on the validation subset showed a global accuracy of 0.932, precision of 0.912 for label "quality" and 0.952 for label "no quality", recall of 0.932 for label "quality" and 0.912 for label "no quality", AUC of 0.980, F1 score of 0.932, and a Matthews correlation coefficient of 0.864. This work offers evidence for the use of topological methods for the process of quality assessment of eye fundus images, where a relatively small vector of characteristics (30 in this case) can enclose enough information for an algorithm to yield classification results useful in the clinical settings of a digital fundoscopy pipeline for CAD.
RESUMO
PURPOSE: On-treatment megavoltage computed tomography on Helical Tomotherapy (Accuray Inc., Sunnyvale, CA) is critical for image guided radiotherapy. A strategy was developed to assess the impact of various jaw widths on image quality and imaging dose with Tomotherapy. METHODS: A cheese phantom (Gammex RMI, Middleton, WI) made of water equivalent materials was employed in this study. Three sets of measurements were independently carried out. Firstly, in the imaging dose measurement, the phantom was placed on the couch and aligned with a stationary green laser and beam isocenter. The measurement point was 10 mm up from the cente of the phantom. Three slices on either side of the middle slice were selected. Secondly, two inserts with different rows of holes of various sizes were placed inside the phantom for image contrast and resolution investigation. Lastly, twelve density inserts were placed into the outer holes in the phantom for measurement of the image value to density table (IVDT). A comparison of imaging dose, image resolution and contrast, IVDT table between different jaw configurations was performed to evaluate the imaging system. RESULTS: Imaging dose was 2.93 cGy with a jaw size of one mm as opposed to 1.62 cGy with a four mm jaw, both of which are below the vendor's requirement: 3 cGy. However, image quality is improved significantly with the smaller jaw. Four lines of holes can be readily identified on images using smaller jaw while only three lines visible with the larger jaw. Image contrast is similarly enhanced when reducing the jaw size. On average CT numbers are 6% higher with the smaller jaw than those obtained with the larger one. CONCLUSIONS: Significant improvement in image quality is achieved with the smaller jaw field in Tomotherapy while the imaging dose is kept at a clinically acceptable level.