Patrick R. Gill

Computational diffractive imager with low-power image change detection

We describe a hardware implementation of an ultra-miniature lensless diffraction-based CMOS computational sensor/imager that supports robust low-power on-sensor image change detection and data streaming modes. Such sensors have numerous applications in surveillance, machine inspection and human interface.

Fast fan/parallel beam CS-based low-dose CT reconstruction

Low dose X-ray Computed Tomography (CT) is clinically desired to reduce the risk of cancer caused by X-ray radiation. Compressed Sensing (CS), which allows images to be formed from incomplete data, enables large dose reduction to be achieved. Though this remains to be clinically unrealized due to excessive computation times. In this paper we demonstrate a fast, complete CS-based ℓ2-TV minimizing CT reconstruction method applicable to both parallel and fan beam geometries to recover high quality images from highly undersampled (thus low-dose) data. We apply the fast pseudo-polar Fourier algorithm and the Central Slice Theorem to reduce the computation time of CS recovery. On a typical desktop computer, we are able to reconstruct a 512×512 CT image in approximately 30 seconds: a clinically-significant speedup compared to the many hours required by previous CS methods.

Computationally efficient Fourier-based image reconstruction in a lensless diffractive imager

We describe Fourier-domain image reconstruction methods for lensless imaging that include two-dimensional signal dewarping to correct optical distortions. These methods offer four orders of magnitude reduction in memory usage than Tikhonov matrix inversion for equivalent image quality.

Accelerated Compressed Sensing Based CT Image Reconstruction

In X-ray computed tomography (CT) an important objective is to reduce the radiation dose without significantly degrading the image quality. Compressed sensing (CS) enables the radiation dose to be reduced by producing diagnostic images from a limited number of projections. However, conventional CS-based algorithms are computationally intensive and time-consuming. We propose a new algorithm that accelerates the CS-based reconstruction by using a fast pseudopolar Fourier based Radon transform and rebinning the diverging fan beams to parallel beams. The reconstruction process is analyzed using a maximum-a-posterior approach, which is transformed into a weighted CS problem. The weights involved in the proposed model are calculated based on the statistical characteristics of the reconstruction process, which is formulated in terms of the measurement noise and rebinning interpolation error. Therefore, the proposed method not only accelerates the reconstruction, but also removes the rebinning and interpolation errors. Simulation results are shown for phantoms and a patient. For example, a 512 × 512 Shepp-Logan phantom when reconstructed from 128 rebinned projections using a conventional CS method had 10% error, whereas with the proposed method the reconstruction error was less than 1%. Moreover, computation times of less than 30 sec were obtained using a standard desktop computer without numerical optimization.

Cholinergic Modulation of Frontoparietal Cortical Network Dynamics Supporting Supramodal Attention

A critical function of attention is to support a state of readiness to enhance stimulus detection, independent of stimulus modality. The nucleus basalis magnocellularis (NBM) is the major source of the neurochemical acetylcholine (ACh) for frontoparietal cortical networks thought to support attention. We examined a potential supramodal role of ACh in a frontoparietal cortical attentional network supporting target detection. We recorded local field potentials (LFPs) in the prelimbic frontal cortex (PFC) and the posterior parietal cortex (PPC) to assess whether ACh contributed to a state of readiness to alert rats to an impending presentation of visual or olfactory targets in one of five locations. Twenty male Long–Evans rats underwent training and then lesions of the NBM using the selective cholinergic immunotoxin 192 IgG-saporin (0.3 μg/μl; ACh-NBM-lesion) to reduce cholinergic afferentation of the cortical mantle. Postsurgery, ACh-NBM-lesioned rats had less correct responses and more omissions than sham-lesioned rats, which changed parametrically as we increased the attentional demands of the task with decreased target duration. This parametric deficit was found equally for both sensory targets. Accurate detection of visual and olfactory targets was associated specifically with increased LFP coherence, in the beta range, between the PFC and PPC, and with increased beta power in the PPC before the targets appearance in sham-lesioned rats. Readiness-associated changes in brain activity and visual and olfactory target detection were attenuated in the ACh-NBM-lesioned group. Accordingly, ACh may support supramodal attention via modulating activity in a frontoparietal cortical network, orchestrating a state of readiness to enhance target detection. SIGNIFICANCE STATEMENT We examined whether the neurochemical acetylcholine (ACh) contributes to a state of readiness for target detection, by engaging frontoparietal cortical attentional networks independent of modality. We show that ACh supported alerting attention to an impending presentation of either visual or olfactory targets. Using local field potentials, enhanced stimulus detection was associated with an anticipatory increase in power in the beta oscillation range before the targets appearance within the posterior parietal cortex (PPC) as well as increased synchrony, also in beta, between the prefrontal cortex and PPC. These readiness-associated changes in brain activity and behavior were attenuated in rats with reduced cortical ACh. Thus, ACh may act, in a supramodal manner, to prepare frontoparietal cortical attentional networks for target detection.

Lensless Smart Sensors: Optical and thermal sensing for the Internet of Things

Lensless Smart Sensors (LSS) add optical and thermal sensing capabilities to the Internet of Things (IoT) in a form factor that cannot be achieved with traditional lensed systems. Different from lensed systems, LSS is based on diffraction instead of refraction, and different from other diffractive optical elements in that it can operate with a wide field of view (FOV) and over a wide wavelength band. LSSs use of computation to extract information from a captured scene makes LSS a good fit for applications where the goal is not to create an image for human consumption, but for machine viewing (e.g. to trigger actions in a connected device). Since the raw sensed image is encoded by the grating structure, LSS opens applications where the use of a camera would create privacy concerns. This paper describes the operational principle of LSS and discusses three examples in more detail.