Gastrointestinal (GI) cancer screening by endoscopy has improved localized cancer prognosis and diagnosis rate. Still, conventional GI endoscopy misses about 8-11% of tumors, due to lack of visibility during upper GI endoscopic exams. One possible way to increase the sensitivity of endoscopic examinations is by using hyperspectral imaging techniques. Hyperspectral imaging captures images across discrete, narrowband wavelength channels, including wavelengths beyond the visible. By analyzing how cells reflect and absorb light across the electromagnetic spectrum, the technique enable users to acquire a unique spectral fingerprint of each cell in a tissue sample. To improve endoscopic imaging and detect cancers at an earlier stage, a team led by professor Baowei Fei at the University of Texas at Dallas (UTD) developed an LED-based, real-time hyperspectral imaging device for endoscopes. The researchers designed a prototype based on a monochrome, micro-digital camera and a multiwavelength LED array comprising 18 LEDs in 18 different wavelengths ranging from 405-910 nm. The team aimed to achieve an image capturing rate of over 10 hypercubes per second (hps) without compromising spatial resolution. The researchers used micro-LEDs with footprints smaller than 400 μm × 400 μm to miniaturize the device. This enabled the team to build an imaging device that could accommodate tens of LEDs at the tip of a clinical endoscopic catheter, and create a hyperspectral imaging system with an in situ light source. By using an in situ hyperspectral light source, the researchers avoided the need for fiber optics, increasing the mobility of the endoscope catheter and lessening the complexity of the mechanical design. The LED-based approach to wavelength scanning makes the device low-power, and allows illumination intensities to be adjusted dynamically based on the distance between the device and the target. University of Texas at Dallas researchers used printed circuit boards to control the developed LED hyperspectral imaging device. Courtesy of the University of Texas at Dallas. To evaluate the feasibility of using an LED-based illumination source for endoscopic imaging, the researchers studied their system’s performance on different normal and cancerous ex vivo tissues. They found that the hyperspectral signatures of different imaging targets acquired using the prototype hyperspectral imaging device were found to be comparable to the data obtained with the reference system. The use of LEDs for hyperspectral imaging could enable numerous applications in endoscopic, laparoscopic, and handheld HSI devices for detecting disease, according to the researchers. Ultimately, Fei aims to develop hyperspectral technology that can be used to track many different types of cancers, and that is small enough to be placed in handheld, affordable personal devices, such as a smartphone or pen that could be used to scan the skin or mouth, for example. “Basically, you could complete the scan, and the information would be wirelessly transferred to the cloud," Fei said. "Then, AI may determine the lesion is suspicious and refer the person to a medical center for follow-up. "Our goal is to produce imaging systems that are really affordable as well as cost-effective, meaning they could find cancers at earlier stages and reduce the need for unnecessary tissue removal and testing.” The research was published in the Journal of Medical Imaging (www.doi.org/10.1117/1.JMI.12.3.035002).