Diagnosing TB with the CellScope

CITRIS researchers have pioneered a new technique that allows clinicians to diagnosis tuberculosis using the cell phone microscope (CellScope). UC Berkeley engineering professors Daniel Fletcher and Jitendra Malik, along with colleagues at UC San Francisco, have developed a way to automatically detect rod-shaped tuberculosis (TB) bacilli using fluorescent images of blood samples through the mobile cell microscope. The method, spearheaded by student Jeannette Chang, will be used as part of upcoming field tests in TB-endemic regions.

TB is the second leading cause of death from infectious disease worldwide, and usually strikes in areas where resources are low, including the doctors, microscopes, and trained technicians necessary to diagnosis the disease. With the CellScope, a low-cost digital microscope, the scientists used computer vision techniques to design an algorithm that can automatically recognize the TB bacilli with high confidence, allowing field workers to effectively treat infected patients. The scientists were able to demonstrate that the new computer vision algorithm performs as well as human technicians.