top of page

RESEARCH PROJECTS

Our Current Focus

kozzi-sound_wave_illustration-2387x1591-

VOICE AS A BIOMARKER FOR DISEASE

We seek to use voice as a marker of disease.  Specifically, laryngeal cancer is one of the few cancers where the morbidity (or disease specific death) is increasing despite advances in treatment protocol.  One of the drivers of the high morbidity in laryngeal cancer is the dearth of early detection options.  As nearly all patients with laryngeal cancer have changes in voice, we seek to use machine learning to identify cancer specific voice changes in a sea of a voice changes from benign processes like the common cold.  We seek advance acoustic analysis and convolutional networks provide a model that will be deployable to edge devices like smartphones.

normal_2016-03-28_vid3_hd 010426 ab.jpeg

OBJECTIVE MEASURES IN LARYNGOSCOPY

We seek to bring objective measures to the very subjective task of laryngeal endoscopy interpretation.  Laryngeal endoscopy, or laryngoscopy, is an critical aspect of the head & neck physical examination.  It is the gold standard for the diagnosis of all laryngeal pathology including laryngeal cancer.  While the task of performing laryngeal endoscopy is straight forward, the interpretation of and ultimate diagnosis from these examinations requires years of dedicated training and expertise.  We seek to democratize and improve the accuracy of laryngeal endoscopy exams but apply machine learning models to endoscopic videos.  We seek to track motion of anatomic structures, estimate the size of lesions using depth estimation techniques and automatically identify pathology using computer vision.

ct-angiography-head-and-neck-2x.jpeg

OBJECTIVE MEASURE IN LARYNGOSCOPY

We seek to bring objective measures to the very subjective task of laryngeal endoscopy interpretation.  Laryngeal endoscopy, or laryngoscopy, is an critical aspect of the head & neck physical examination.  It is the gold standard for the diagnosis of all laryngeal pathology including laryngeal cancer.  While the task of performing laryngeal endoscopy is straight forward, the interpretation of and ultimate diagnosis from these examinations requires years of dedicated training and expertise.  We seek to democratize and improve the accuracy of laryngeal endoscopy exams but apply machine learning models to endoscopic videos.  We seek to track motion of anatomic structures, estimate the size of lesions using depth estimation techniques and automatically identify pathology using computer vision.

Embryonic Stem Cells

OBJECTIVE MEASURES IN LARYNGOSCOPY

We seek to bring objective measures to the very subjective task of laryngeal endoscopy interpretation.  Laryngeal endoscopy, or laryngoscopy, is an critical aspect of the head & neck physical examination.  It is the gold standard for the diagnosis of all laryngeal pathology including laryngeal cancer.  While the task of performing laryngeal endoscopy is straight forward, the interpretation of and ultimate diagnosis from these examinations requires years of dedicated training and expertise.  We seek to democratize and improve the accuracy of laryngeal endoscopy exams but apply machine learning models to endoscopic videos.  We seek to track motion of anatomic structures, estimate the size of lesions using depth estimation techniques and automatically identify pathology using computer vision.

Research: Research
bottom of page