Combined analysis of sMRI and fMRI imaging data provides accurate disease markers for hearing impairment

Abstract

In this research, we developed a robust two-layer classifier that can accurately classify normal hearing (NH) from hearing impaired (HI) infants with congenital sensori-neural hearing loss (SNHL) based on their Magnetic Resonance (MR) images. Unlike traditional methods that examine the intensity of each single voxel, we extracted high-level features to characterize the structural MR images (sMRI) and functional MR images (fMRI). The Scale Invariant Feature Transform (SIFT) algorithm was employed to detect and describe the local features in sMRI. For fMRI, we constructed contrast maps and detected the most activated/de-activated regions in each individual. Based on those salient regions occurring across individuals, the bag-of-words strategy was introduced to vectorize the contrast maps. We then used a two-layer model to integrate these two types of features together. With the leave-one-out cross-validation approach, this integrated model achieved an AUC score of 0.90. Additionally, our algorithm highlighted several important brain regions that differentiated between NH and HI children. Some of these regions, e.g. planum temporale and angular gyrus, were well known auditory and visual language association regions. Others, e.g. the anterior cingulate cortex (ACC), were not necessarily expected to play a role in differentiating HI from NH children and provided a new understanding of brain function and of the disorder itself. These important brain regions provided clues about neuroimaging markers that may be relevant to the future use of functional neuroimaging to guide predictions about speech and language outcomes in HI infants who receive a cochlear implant. This type of prognostic information could be extremely useful and is currently not available to clinicians by any other means.

fMRI as a Preimplant Objective Tool to Predict Postimplant Oral Language Outcomes in Children with Cochlear Implants

“…The presence of the aforementioned clusters is consistent with the early engagement of this "ventral stream" before 24 months old in infants with hearing loss. Using the same auditory stimulation paradigm as in the present study, the angular gyrus, the cingulate gyrus, and regions in the prefrontal cortex were identified to be effective classifiers of normal versus impaired hearing in 39 children using a machine learning technique (Tan et al 2013). The present study takes this finding one step further by investigating the role of these regions in postimplant oral language performance in CI recipients.…”

“…Each segment lasted for 5 seconds. This natural and ecologically plausible stimulus is one that we have used extensively in fMRI studies, including young children Sroka et al 2015) and infants (DiFrancesco et al 2013;Tan et al 2013) and consistently demonstrates robust bilateral auditory and language activation, even under the influence of sleep (Wilke et al 2003) and anesthesia (DiFrancesco et al 2013). Given its demonstrated utility in producing strong brain activation in auditory language circuitry in the brain, we chose to use it in this study.…”

“…Auditory stimuli were presented during silent periods of sparse echo planar imaging (EPI-BOLD) acquisition for a period of 5 seconds and the scanner acquired images between the auditory stimulus conditions for a period of 6 seconds. Each auditory condition was preceded by an additional 1 second of silence to allow for decay of the hemodynamic BOLD response to the scanner noise (e.g., Tan et al 2013). One block of stimulus acquisition is outlined in Figure 1.…”

“…Previous studies (Patel et al 2007;Smith et al 2011;Tan et al 2013) have established feasibility of using the current auditory stimulation protocol in children with hearing loss. Nonetheless, the effects of acoustic distortion resulting due to presentation of high-intensity auditory stimuli through headphones cannot be ruled out.…”

“…That work demonstrated the potential of fMRI as a prognostic indicator of postimplant auditory outcomes. Recently, Tan et al (2013) used a machine learning approach to classify infants with and without hearing loss based on their structural and functional neuroimaging data. They used the support vector machine algorithm (e.g., Orrù et al 2012) to identify key regions such as the angular gyrus, the cingulate gyrus, and regions in the prefrontal cortex that effectively distinguished between the brains of the two groups.…”