Large-scale annotated dataset for cochlear hair cell detection and classification

Our sense of hearing is mediated by cochlear hair cells, of which there are two types organized in one row of inner hair cells and three rows of outer hair cells. Each cochlea contains 5-15 thousand terminally differentiated hair cells, and their survival is essential for hearing as they do not regenerate after insult. It is often desirable in hearing research to quantify the number of hair cells within cochlear samples, in both pathological conditions, and in response to treatment. Machine learning can be used to automate the quantification process but requires a vast and diverse dataset for effective training. In this study, we present a large collection of annotated cochlear hair-cell datasets, labeled with commonly used hair-cell markers and imaged using various fluorescence microscopy techniques. The collection includes samples from mouse, rat, guinea pig, pig, primate, and human cochlear tissue, from normal conditions and following in-vivo and in-vitro ototoxic drug application. The dataset includes over 107,000 hair cells which have been identified and annotated as either inner or outer hair cells. This dataset is the result of a collaborative effort from multiple laboratories and has been carefully curated to represent a variety of imaging techniques. With suggested usage parameters and a well-described annotation procedure, this collection can facilitate the development of generalizable cochlear hair-cell detection models or serve as a starting point for fine-tuning models for other analysis tasks. By providing this dataset, we aim to give other hearing research groups the opportunity to develop their own tools with which to analyze cochlear imaging data more fully, accurately, and with greater ease.

PKHD1L1 is required for stereocilia bundle maintenance, durable hearing function and resilience to noise exposure

Sensory hair cells of the cochlea are essential for hearing, relying on the mechanosensitive stereocilia bundle at their apical pole for their function. Polycystic Kidney and Hepatic Disease 1-Like 1 (PKHD1L1) is a stereocilia protein required for normal hearing in mice, and for the formation of the transient stereocilia surface coat, expressed during early postnatal development. While the function of the stereocilia coat remains unclear, growing evidence supports PKHD1L1 as a human deafness gene. In this study we carry out in depth characterization of PKHD1L1 expression in mice during development and adulthood, analyze hair-cell bundle morphology and hearing function in aging PKHD1L1-defficient mouse lines, and assess their susceptibility to noise damage. Our findings reveal that PKHD1L1-deficient mice display no disruption to bundle cohesion or tectorial membrane attachment-crown formation during development. However, starting from 6 weeks of age, PKHD1L1-defficient mice display missing stereocilia and disruptions to bundle coherence. Both conditional and constitutive PKHD1L1 knock-out mice develop high-frequency hearing loss progressing to lower frequencies with age. Furthermore, PKHD1L1-deficient mice are susceptible to permanent hearing loss following moderate acoustic overexposure, which induces only temporary hearing threshold shifts in wild-type mice. These results suggest a role for PKHD1L1 in establishing robust sensory hair bundles during development, necessary for maintaining bundle cohesion and function in response to acoustic trauma and aging.

Large-scale annotated dataset for cochlear hair cell detection and classification

Our sense of hearing is mediated by cochlear hair cells, localized within the sensory epithelium called the organ of Corti. There are two types of hair cells in the cochlea, which are organized in one row of inner hair cells and three rows of outer hair cells. Each cochlea contains a few thousands of hair cells, and their survival is essential for our perception of sound because they are terminally differentiated and do not regenerate after insult. It is often desirable in hearing research to quantify the number of hair cells within cochlear samples, in both pathological conditions, and in response to treatment. However, the sheer number of cells along the cochlea makes manual quantification impractical. Machine learning can be used to overcome this challenge by automating the quantification process but requires a vast and diverse dataset for effective training. In this study, we present a large collection of annotated cochlear hair-cell datasets, labeled with commonly used hair-cell markers and imaged using various fluorescence microscopy techniques. The collection includes samples from mouse, human, pig and guinea pig cochlear tissue, from normal conditions and following in-vivo and in-vitro ototoxic drug application. The dataset includes over 90'000 hair cells, all of which have been manually identified and annotated as one of two cell types: inner hair cells and outer hair cells. This dataset is the result of a collaborative effort from multiple laboratories and has been carefully curated to represent a variety of imaging techniques. With suggested usage parameters and a well-described annotation procedure, this collection can facilitate the development of generalizable cochlear hair cell detection models or serve as a starting point for fine-tuning models for other analysis tasks. By providing this dataset, we aim to supply other groups within the hearing research community with the opportunity to develop their own tools with which to analyze cochlear imaging data more fully, accurately, and with greater ease.

SKOOTS: Skeleton oriented object segmentation for mitochondria

The segmentation of individual instances of mitochondria from imaging datasets is informative, yet time-consuming to do by hand, sparking interest in developing automated algorithms using deep neural networks. Existing solutions for various segmentation tasks are largely optimized for one of two types of biomedical imaging: high resolution three-dimensional (whole neuron segmentation in volumetric electron microscopy datasets) or two-dimensional low resolution (whole cell segmentation of light microscopy images). The former requires consistently predictable boundaries to segment large structures, while the latter is boundary invariant but struggles with segmentation of large 3D objects without downscaling. Mitochondria in whole cell 3D EM datasets often occupy the challenging middle ground: large with ambiguous borders, limiting accuracy with existing tools. To rectify this, we have developed skeleton oriented object segmentation (SKOOTS); a new segmentation approach which efficiently handles large, densely packed mitochondria. We show that SKOOTS can accurately, and efficiently, segment 3D mitochondria in previously difficult situations. Furthermore, we will release a new, manually annotated, 3D mitochondria segmentation dataset. Finally, we show this approach can be extended to segment objects in 3D light microscopy datasets. These results bridge the gap between existing segmentation approaches and increases the accessibility for three-dimensional biomedical image analysis.

The hair cell analysis toolbox is a precise and fully automated pipeline for whole cochlea hair cell quantification

Our sense of hearing is mediated by sensory hair cells, precisely arranged and highly specialized cells subdivided into outer hair cells (OHCs) and inner hair cells (IHCs). Light 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. 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. Here, we present a highly accurate machine learning-based hair cell analysis toolbox (HCAT) for the comprehensive analysis of whole cochleae (or smaller regions of interest) across light microscopy imaging modalities and species. The HCAT is a software that automates common image analysis tasks such as counting hair cells, classifying them by subtype (IHCs versus 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.

Identification of a series of hair-cell MET channel blockers that protect against aminoglycoside-induced ototoxicity

To identify small molecules that shield mammalian sensory hair cells from the ototoxic side effects of aminoglycoside antibiotics, 10,240 compounds were initially screened in zebrafish larvae, selecting for those that protected lateral-line hair cells against neomycin and gentamicin. When the 64 hits from this screen were retested in mouse cochlear cultures, 8 protected outer hair cells (OHCs) from gentamicin in vitro without causing hair-bundle damage. These 8 hits shared structural features and blocked, to varying degrees, the OHC's mechano-electrical transducer (MET) channel, a route of aminoglycoside entry into hair cells. Further characterization of one of the strongest MET channel blockers, UoS-7692, revealed it additionally protected against kanamycin and tobramycin and did not abrogate the bactericidal activity of gentamicin. UoS-7692 behaved, like the aminoglycosides, as a permeant blocker of the MET channel; significantly reduced gentamicin-Texas red loading into OHCs; and preserved lateral-line function in neomycin-treated zebrafish. Transtympanic injection of UoS-7692 protected mouse OHCs from furosemide/kanamycin exposure in vivo and partially preserved hearing. The results confirmed the hair-cell MET channel as a viable target for the identification of compounds that protect the cochlea from aminoglycosides and provide a series of hit compounds that will inform the design of future otoprotectants.

Accelerated Age-Related Degradation of the Tectorial Membrane in the Ceacam16βgal/βgal Null Mutant Mouse, a Model for Late-Onset Human Hereditary Deafness DFNB113

CEACAM16 is a non-collagenous protein of the tectorial membrane, an extracellular structure of the cochlea essential for normal hearing. Dominant and recessive mutations in CEACAM16 have been reported to cause postlingual and progressive forms of deafness in humans. In a previous study of young Ceacam16βgal/βgal null mutant mice on a C57Bl/6J background, the incidence of spontaneous otoacoustic emissions (SOAEs) was greatly increased relative to Ceacam16+/+ and Ceacam16+/βgal mice, but auditory brain-stem responses (ABRs) and distortion product otoacoustic emissions (DPOAEs) were near normal, indicating auditory thresholds were not significantly affected. To determine if the loss of CEACAM16 leads to hearing loss at later ages in this mouse line, cochlear structure and auditory function were examined in Ceacam16+/+, Ceacam16+/βgal and Ceacam16βgal/βgal mice at 6 and 12 months of age and compared to that previously described at 1 month. Analysis of older Ceacam16βgal/βgal mice reveals a progressive loss of matrix from the core of the tectorial membrane that is more extensive in the apical, low-frequency regions of the cochlea. In Ceacam16βgal/βgal mice at 6-7 months, the DPOAE magnitude at 2f1-f2 and the incidence of SOAEs both decrease relative to young animals. By ∼12 months, SOAEs and DPOAEs are not detected in Ceacam16βgal/βgal mice and ABR thresholds are increased by up to ∼40 dB across frequency, despite a complement of hair cells similar to that present in Ceacam16+/+ mice. Although SOAE incidence decreases with age in Ceacam16βgal/βgal mice, it increases in aging heterozygous Ceacam16+/βgal mice and is accompanied by a reduction in the accumulation of CEACAM16 in the tectorial membrane relative to controls. An apically-biased loss of matrix from the core of the tectorial membrane, similar to that observed in young Ceacam16βgal/βgal mice, is also seen in Ceacam16+/+ and Ceacam16+/βgal mice, and other strains of wild-type mice, but at much later ages. The loss of Ceacam16 therefore accelerates age-related degeneration of the tectorial membrane leading, as in humans with mutations in CEACAM16, to a late-onset progressive form of hearing loss.