Our project's research goal was to develop an objective visual assessment of ADSD that correlates with subjective voice analyses of ADSD patients.
Brief - Methods, Development, Results, and Impact
I developed and tested the computer vision pipeline.
Image segmentation is done with pixel thresholding, which is optimized by a neural network which predicts an image's optimal luminance threshold. Then, another neural network predicts which binary object represents the glottis.
The figure to the right depicts the regression models.
Our results are significant and hold value in further understanding ADSD and how characteristics of the disorder relate with voice quality.
Previous Solution Attempts
CNN Optimization Architecture
I built a convolutional neural network (CNN) using MATLAB to predict patients’ ADSD severity based off of images of their larynx.
This involved the design and implementation of an optimization architecture for testing parameters core to the internals of the CNN, as well as parameters regarding training/testing methodology.