Distorted image classification using neural activation pattern matching loss.

In image classification, a deep neural network (DNN) that is trained on undistorted images constitutes an effective decision boundary. Unfortunately, this boundary does not support distorted images, such as noisy or blurry ones, leading to accuracy drop-off. As a simple approach for classifying distorted images as well as undistorted ones, previous methods have optimized the trained DNN again on both kinds of images. However, in these methods, the decision boundary may become overly complicated during optimization because there is no regularization of the decision boundary. Consequently, this decision boundary limits efficient optimization. In this paper, we study a simple yet effective decision boundary for distorted image classification through the use of a novel loss, called a "neural activation pattern matching (NAPM) loss". The NAPM loss is based on recent findings that the decision boundary is a piecewise linear function, where each linear segment is constructed from a neural activation pattern in the DNN when an image is fed to it. The NAPM loss extracts the neural activation patterns when the distorted image and its undistorted version are fed to the DNN and then matches them with each other via the sigmoid cross-entropy. Therefore, it constrains the DNN to classify the distorted image and its undistorted version by the same linear segment. As a result, our loss accelerates efficient optimization by preventing the decision boundary from becoming overly complicated. Our experiments demonstrate that our loss increases the accuracy of the previous methods in all conditions evaluated.

Liposome-encapsulated hemoglobin alleviates hearing loss after transient cochlear ischemia and reperfusion in the gerbil.

To test liposome-encapsulated hemoglobin (LEH) in transient cochlear ischemia/reperfusion as a model of sudden deafness, Mongolian gerbils were randomly assigned to receive 2 mL/kg of either low-affinity LEH (l-LEH, P5002 = 40 mm Hg), high-affinity LEH (h-LEH, P5002 = 10 mm Hg), homologous red blood cells (RBCs), or saline (each group n = 6) 30 min before 15-min occlusion of the bilateral vertebral arteries and reperfusion. Sequential changes in hearing were assessed by auditory brain response 1, 4, and 7 days after ischemia/reperfusion, when the animals were sacrificed for pathological studies. h-LEH was significantly more protective than l-LEH in suppressing hearing loss, in contrast to RBC or saline treatment, at 8, 16, and 32 kHz, where hearing loss was most severe (P < 0.05 between any two groups) on the first day after cochlear ischemia/reperfusion. Thereafter, hearing loss improved gradually in all groups, with a significant difference among groups up to 7 days, when morphological studies revealed that the inner hair cells but not the outer hair cells, were significantly lost in the groups in the same order. The results suggest that pretreatment with h-LEH is significantly more protective than l-LEH in mitigating hearing loss and underlying pathological damage, in contrast to transfusion or saline infusion 7 days after transient cochlear ischemia/reperfusion.

Delayed neuronal cell death in brainstem after transient brainstem ischemia in gerbils.

BACKGROUND: Because of the lack of reproducible brainstem ischemia models in rodents, the temporal profile of ischemic lesions in the brainstem after transient brainstem ischemia has not been evaluated intensively. Previously, we produced a reproducible brainstem ischemia model of Mongolian gerbils. Here, we showed the temporal profile of ischemic lesions after transient brainstem ischemia. RESULTS: Brainstem ischemia was produced by occlusion of the bilateral vertebral arteries just before their entry into the transverse foramina of the cervical vertebrae of Mongolian gerbils. Animals were subjected to brainstem ischemia for 15 min, and then reperfused for 0 d (just after ischemia), 1 d, 3 d and 7 d (n = 4 in each group). Sham-operated animals (n = 4) were used as control. After deep anesthesia, the gerbils were perfused with fixative for immunohistochemical investigation. Ischemic lesions were detected by immunostaining for microtubule-associated protein 2 (MAP2). Just after 15-min brainstem ischemia, ischemic lesions were detected in the lateral vestibular nucleus and the ventral part of the spinal trigeminal nucleus, and these ischemic lesions disappeared one day after reperfusion in all animals examined. However, 3 days and 7 days after reperfusion, ischemic lesions appeared again and clusters of ionized calcium-binding adapter molecule-1(IBA-1)-positive cells were detected in the same areas in all animals. CONCLUSION: These results suggest that delayed neuronal cell death took place in the brainstem after transient brainstem ischemia in gerbils.

Local hypothermia in the treatment of idiopathic sudden sensorineural hearing loss.

OBJECTIVE: The additive effects of local hypothermia and restricted activity in the treatment of idiopathic sudden sensorineural hearing loss (ISSHL) were investigated by case-matched study as a multicenter (13 hospitals) pilot trial. PATIENTS AND METHODS: In a preliminary experiment, we evaluated the effects of cooled water pillow (15 degrees C). Cooling the neck and mastoid with the pillow decreased the tympanic membrane temperature for 1.4 degrees C in 2h without causing uncomfortable sensation or frostbite. In this study, 86 patients with ISSHL were enrolled in the hypothermic group, which received hypothermic treatment with restricted activity in addition to medication, and 86 ISSHL patients constituted the control group, which received the same medication but without cooling and rest. Control patients were selected retrospectively from case records by matching the experimental patients with respect to age, gender, days until the start of treatment, hearing loss, shape of the audiogram, and accompanying vertigo. The patients in the hypothermic group were admitted and treated with a cooled water pillow for 48h, in addition to conventional drug treatment (e.g., 60 mg of prednisone) for 7 days. The water pillow was cooled to 15 degrees C and was changed 4-5 times per day. The patients used the water pillow for the first 48 h after admission, with restricted activity. The control patients received only the medications. RESULTS: Hearing results were evaluated using criteria proposed by the Sudden Sensorineural Hearing Loss Research Group of the Japanese Ministry of Health and Welfare. The recovery rates were judged 6 months after onset. The recovery rate in the hypothermic group was significantly (p<0.05) better than that in the control group. When the comparison was limited to younger patients, the use of the cooled water pillow was effective in facilitating the recovery of hearing. CONCLUSIONS: Hearing restoration in ISSHL may be improved by adding mild hypothermia and restricted activity to the conventional treatment.

Prednisolone prevents transient ischemia-induced cochlear damage in gerbils.

CONCLUSIONS: Prednisolone protects against inner ear damage, even when administered after ischemic injury in Mongolian gerbils. OBJECTIVE: The effect of prednisolone on ischemia-induced cochlear damage was investigated in Mongolian gerbils. MATERIALS AND METHODS: The bilateral vertebral arteries were occluded for 15 min to transiently induce cochlear ischemia, followed by an intraperitoneal injection of prednisolone (1 mg/kg) or physiological saline (control). Sequential changes in hearing were evaluated by recording the auditory brainstem response (ABR) before and at 1, 4, and 7 days after treatment. In our histologic analysis, the numbers of dead and intact inner hair cells (IHCs) were counted in specimens stained with rhodamine-phalloidin. RESULTS: In control animals, transient ischemia increased the ABR threshold (24.2 +/- 8.6 dB) within 7 days of treatment, whereas prednisolone-treated animals exhibited a threshold of 14.2 +/- 9.2 dB. Furthermore, the percent IHC loss at the basal turn of the cochlea was 26.5 +/- 11.4% in control animals compared with 5.3 +/- 3.0% in the prednisolone-treated group.

Ischemic tolerance in the cochlea.

Ischemic tolerance in the cochlea was investigated in a gerbil model of cochlear ischemia. Transient cochlear ischemia was produced by extracranial occlusion of the bilateral vertebral arteries. The gerbils were divided into two groups; single ischemia group and double ischemia group. In the single ischemia group, animals were subjected to lethal cochlear ischemia for 15 min. In the double ischemia group, animals were subjected to sublethal cochlear ischemia for 2 min at 2 days before lethal ischemia for 15 min. Consequently, sublethal ischemia prevented lethal ischemia-induced hair cell degeneration and ameliorated hearing impairment, suggesting ischemic tolerance in the cochlea.

Insulin-like growth factor 1 treatment via hydrogels rescues cochlear hair cells from ischemic injury.

This study was designed to investigate the protective effects of recombinant human insulin-like growth factor 1 (rhIGF1), applied locally via a hydrogel, against ischemic damage of the cochleae in gerbils. A hydrogel was immersed in rhIGF1 or saline and was applied on the round window membrane 30 min after the ischemia. Local rhIGF1 treatment significantly reduced the elevation of auditory brain responses thresholds at a frequency of 8 kHz on days 1, 4, and 7 after ischemia. A histological analysis revealed increased survival of inner hair cells in the animals treated with rhIGF1 via the hydrogel 7 days after ischemia. These findings showed that local rhIGF1 application using a hydrogel has the potential to protect the cochleae from ischemic injury.

Postischemic mild hypothermia alleviates hearing loss because of transient ischemia.

The effect of postischemic mild hypothermia on the inner ear has not been clarified. In this study, we investigated whether hypothermia after transient ischemia could prevent cochlear damage and its therapeutic time window. Mongolian gerbils were divided into six groups: a sham-operation group, a normothermia group, and four hypothermia groups in which hypothermia was induced 1-7, 1-4, 3-6, and 6-9 h after reperfusion. Animals subjected to postischemic mild hypothermia within 3 h after reperfusion had attenuated hearing loss and inner hair cell loss. The protective effect was greater when hypothermia was induced earlier and had a longer duration. This implies that mild hypothermia after ischemia could have therapeutic effects for inner ear ischemic damage.