Noise-trained deep neural networks effectively predict human vision and its neural responses to challenging images.

Deep neural networks (DNNs) for object classification have been argued to provide the most promising model of the visual system, accompanied by claims that they have attained or even surpassed human-level performance. Here, we evaluated whether DNNs provide a viable model of human vision when tested with challenging noisy images of objects, sometimes presented at the very limits of visibility. We show that popular state-of-the-art DNNs perform in a qualitatively different manner than humans-they are unusually susceptible to spatially uncorrelated white noise and less impaired by spatially correlated noise. We implemented a noise training procedure to determine whether noise-trained DNNs exhibit more robust responses that better match human behavioral and neural performance. We found that noise-trained DNNs provide a better qualitative match to human performance; moreover, they reliably predict human recognition thresholds on an image-by-image basis. Functional neuroimaging revealed that noise-trained DNNs provide a better correspondence to the pattern-specific neural representations found in both early visual areas and high-level object areas. A layer-specific analysis of the DNNs indicated that noise training led to broad-ranging modifications throughout the network, with greater benefits of noise robustness accruing in progressively higher layers. Our findings demonstrate that noise-trained DNNs provide a viable model to account for human behavioral and neural responses to objects in challenging noisy viewing conditions. Further, they suggest that robustness to noise may be acquired through a process of visual learning.

Characterizing the effects of feature salience and top-down attention in the early visual system.

The visual system employs a sophisticated balance of attentional mechanisms: salient stimuli are prioritized for visual processing, yet observers can also ignore such stimuli when their goals require directing attention elsewhere. A powerful determinant of visual salience is local feature contrast: if a local region differs from its immediate surround along one or more feature dimensions, it will appear more salient. We used high-resolution functional MRI (fMRI) at 7T to characterize the modulatory effects of bottom-up salience and top-down voluntary attention within multiple sites along the early visual pathway, including visual areas V1-V4 and the lateral geniculate nucleus (LGN). Observers viewed arrays of spatially distributed gratings, where one of the gratings immediately to the left or right of fixation differed from all other items in orientation or motion direction, making it salient. To investigate the effects of directed attention, observers were cued to attend to the grating to the left or right of fixation, which was either salient or nonsalient. Results revealed reliable additive effects of top-down attention and stimulus-driven salience throughout visual areas V1-hV4. In comparison, the LGN exhibited significant attentional enhancement but was not reliably modulated by orientation- or motion-defined salience. Our findings indicate that top-down effects of spatial attention can influence visual processing at the earliest possible site along the visual pathway, including the LGN, whereas the processing of orientation- and motion-driven salience primarily involves feature-selective interactions that take place in early cortical visual areas.NEW & NOTEWORTHY While spatial attention allows for specific, goal-driven enhancement of stimuli, salient items outside of the current focus of attention must also be prioritized. We used 7T fMRI to compare salience and spatial attentional enhancement along the early visual hierarchy. We report additive effects of attention and bottom-up salience in early visual areas, suggesting that salience enhancement is not contingent on the observer's attentional state.

Copolymer-Mediated Cell Aggregation Promotes a Proangiogenic Stem Cell Phenotype In Vitro and In Vivo.

Material-induced cell aggregation drives a proangiogenic expression profile. Copolymer substrates containing cell-repellent and cell-adhesive domains force the aggregation of human mesenchymal stem cells, which results in enhanced tubulogenesis in vitro and stabilization of vasculature in vivo. These findings can be used to design instructive biomaterial scaffolds for clinical use.

In vivo hyperspectral imaging of microvessel response to trastuzumab treatment in breast cancer xenografts.

HER2-amplified (HER2 + ) breast cancers are treated with the anti-HER2 monoclonal antibody trastuzumab. Although trastuzumab reduces production of the angiogenic factor VEGF in HER2 + tumors, the acute and sustained effects of trastuzumab on the tumor vasculature are not understood fully, particularly in trastuzumab-resistant tumors. We used mouse models of trastuzumab sensitive and trastuzumab-resistant HER2 + breast cancers to measure dynamic changes in tumor microvessel density and hemoglobin oxygenation (sO2) in vivo using quantitative hyperspectral imaging at 2, 5, 9, and 14 days after antibody treatment. Further analysis quantified the distribution of microvessels into low and high oxygenation groups, and monitored changes in these distributions with trastuzumab treatment. Gold standard immunohistochemistry was performed to validate complementary markers of tumor cell and vascular response to treatment. Trastuzumab treatment in both responsive and resistant tumors resulted in decreased sO2 5 days after initial treatment when compared to IgG-treated controls (p<0.05). Importantly, responsive tumors showed significantly higher vessel density and significantly lower sO2 than all other groups at 5 days post-treatment (p<0.05). Distribution analysis of vessel sO2 showed a significant (p<0.05) shift of highly oxygenated vessels towards lower oxygenation over the time-course in both trastuzumab-treated responsive and resistant tumors. This study suggests that longitudinal hyperspectral imaging of microvessel sO2 and density could distinguish trastuzumab-responsive from trastuzumab-resistant tumors, a finding that could be exploited in the post-neoadjuvant setting to guide post-surgical treatment decisions.

Quantifying the vascular response to ischemia with speckle variance optical coherence tomography.

Longitudinal monitoring techniques for preclinical models of vascular remodeling are critical to the development of new therapies for pathological conditions such as ischemia and cancer. In models of skeletal muscle ischemia in particular, there is a lack of quantitative, non-invasive and long term assessment of vessel morphology. Here, we have applied speckle variance optical coherence tomography (OCT) methods to quantitatively assess vascular remodeling and growth in a mouse model of peripheral arterial disease. This approach was validated on two different mouse strains known to have disparate rates and abilities of recovering following induction of hind limb ischemia. These results establish the potential for speckle variance OCT as a tool for quantitative, preclinical screening of pro- and anti-angiogenic therapies.

Enhanced photoacoustic detection of melanoma cells using gold nanoparticles.

BACKGROUND AND OBJECTIVES: We tagged melanoma cells with gold nanoparticles to show their viability for increasing sensitivity in a photoacoustic detection system. Ultimately, this study models the detection of circulating tumor cells, which are an important prognostic factor in the progress of melanoma. STUDY DESIGN/MATERIALS AND METHODS: A Q-switched, tunable Nd:YAG laser was used to irradiate cells in both a stationary and flow set-up. Photoacoustic signals were measured using a polyvinylidene fluoride (PVDF) film in the stationary test, and a commercially available ultrasonic probe for flow tests. Both unmodified melanoma cells and gold nanoparticle (AuNP) tagged melanoma were tested. RESULTS: AuNP tagged melanoma in a stationary set-up showed an average of 0.227 mV/mJ larger signal than the untagged, indicating a signal increase of 34%. At 500 nm there is a maximum difference of 0.295 mV/mJ, or a 41% increase. In flow tests, the ultrasound probe was able to detect single cells, but the increased signal from AuNP tagging was minimal. CONCLUSION: AuNP tagging proved to give an increased photoacoustic signal allowing greater sensitivity in stationary metastasized melanoma detection systems using photoacoustics.

Photoacoustic detection of melanoma micrometastasis in sentinel lymph nodes.

Melanoma is the deadliest form of skin cancer and has the fastest growth rate of all cancer types. Proper staging of melanoma is required for clinical management. One method of staging melanoma is performed by taking a sentinel node biopsy, in which the first node in the lymphatic drainage path of the primary lesion is removed and tested for the presence of melanoma cells. Current standard of care typically involves taking fewer than ten histologic sections of the node out of the hundreds of possible sections available in the tissue. We have developed a photoacoustic method that probes the entire intact node. We acquired a lymph node from a healthy canine subject. We cultured a malignant human melanoma cell line HS 936. Approximately 1 x 10(6) cells were separated and injected into the lymph node. We also had a healthy lymph node in which no melanoma cells were implanted. We used a tunable laser system set at 532 nm to irradiate the lymph nodes. Three piezoelectric acoustic detectors were positioned near the lymph node to detect photoacoustic pulses generated within the lymph nodes. We also acquired lymph nodes from pigs and repeated the experiments with increased amplification and improved sensors. We detected photoacoustic responses from a lymph node with as few as 500 melanoma cells injected into the tissue, while normal lymph nodes showed no response. Photoacoustic generation can be used to detect melanoma micrometastasis in sentinel lymph nodes. This detection can be used to guide further histologic study of the node, increasing the accuracy of the sentinel lymph node biopsy.