| Popis: |
Our sense of hearing is mediated by sensory hair cells, precisely arranged and highly specialized cells subdivided into two subtypes: outer hair cells (OHCs) which amplify sound-induced mechanical vibration, and inner hair cells (IHCs) which convert vibrations into electrical signals for interpretation by the brain. One row of IHCs and three rows of OHCs are arranged tonotopically; cells at a particular location respond best to a specific frequency which decreases from base to apex of the cochlea. Loss of hair cells at a specific place affects hearing performance at the corresponding tonotopic frequency. To better understand the underlying cause of hearing loss in patients (or experimental animals) a plot of hair cell survival along the cochlear frequency map, known as a cochleogram, can be generated post-mortem, involving manually counting thousands of cells. Currently, there are no widely applicable tools for fast, unsupervised, unbiased, and comprehensive image analysis of auditory hair cells that work well either with imaging datasets containing an entire cochlea or smaller sampled regions. Current microscopy tools allow for imaging of auditory hair cells along the full length of the cochlea, often yielding more data than feasible to manually analyze. Here, we present a machine learning-based hair cell analysis toolbox for the comprehensive analysis of whole cochleae (or smaller regions of interest). The Hair Cell Analysis Toolbox (HCAT) is a software that automates common image analysis tasks such as counting hair cells, classifying them by subtype (IHCs vs OHCs), determining their best frequency based on their location along the cochlea, and generating cochleograms. These automated tools remove a considerable barrier in cochlear image analysis, allowing for faster, unbiased, and more comprehensive data analysis practices. Furthermore, HCAT can serve as a template for deep-learning-based detection tasks in other types of biological tissue: with some training data, HCAT’s core codebase can be trained to develop a custom deep learning detection model for any object on an image. |
