Date of Award

Summer 8-15-2018

Author's Department

Electrical & Systems Engineering

Degree Name

Doctor of Philosophy (PhD)

Degree Type



Background: Cancer remains a major public health problem worldwide and poses a huge economic burden. Near-infrared (NIR) fluorescence image-guided surgery (IGS) utilizes molecular markers and imaging instruments to identify and locate tumors during surgical resection. Unfortunately, current state-of-the-art NIR fluorescence imaging systems are bulky, costly, and lack both fluorescence sensitivity under surgical illumination and co-registration accuracy between multimodal images. Additionally, the monitor-based display units are disruptive to the surgical workflow and are suboptimal at indicating the 3-dimensional position of labeled tumors. These major obstacles have prevented the wide acceptance of NIR fluorescence imaging as the standard of care for cancer surgery. The goal of this dissertation is to enhance cancer treatment by developing novel image sensors and presenting the information using holographic augmented reality (AR) display to the physician in intraoperative settings.

Method: By mimicking the visual system of the Morpho butterfly, several single-chip, color-NIR fluorescence image sensors and systems were developed with CMOS technologies and pixelated interference filters. Using a holographic AR goggle platform, an NIR fluorescence IGS display system was developed. Optoelectronic evaluation was performed on the prototypes to evaluate the performance of each component, and small animal models and large animal models were used to verify the overall effectiveness of the integrated systems at cancer detection.

Result: The single-chip bio-inspired multispectral logarithmic image sensor I developed has better main performance indicators than the state-of-the-art NIR fluorescence imaging instruments. The image sensors achieve up to 140 dB dynamic range. The sensitivity under surgical illumination achieves 6108 V/(mW/cm2), which is up to 25 times higher. The signal-to-noise ratio is up to 56 dB, which is 11 dB greater. These enable high sensitivity fluorescence imaging under surgical illumination. The pixelated interference filters enable temperature-independent co-registration accuracy between multimodal images. Pre-clinical trials with small animal model demonstrate that the sensor can achieve up to 95% sensitivity and 94% specificity with tumor-targeted NIR molecular probes. The holographic AR goggle provides the physician with a non-disruptive 3-dimensional display in the clinical setup. This is the first display system that co-registers a virtual image with human eyes and allows video rate image transmission. The imaging system is tested in the veterinary science operating room on canine patients with naturally occurring cancers. In addition, a time domain pulse-width-modulation address-event-representation multispectral image sensor and a handheld multispectral camera prototype are developed.

Conclusion: The major problems of current state-of-the-art NIR fluorescence imaging systems are successfully solved. Due to enhanced performance and user experience, the bio-inspired sensors and augmented reality display system will give medical care providers much needed technology to enable more accurate value-based healthcare.


English (en)


Shantanu Chakrabartty

Committee Members

Martin Arthur, Shantanu Chakrabartty, Roger Chamberlain, Viktor Gruev,


Permanent URL: https://doi.org/10.7936/28s2-nx53