Early Diagnosis and Treatment of Tuberculosis
Early Diagnosis and Treatment of Tuberculosis
From HTN
Contents |
Challenge Overview
TB is a curable disease given the existence of effective drugs for treatment. However, the chance of cure is reduced when treatment starts at advanced stages of the disease and when the infection is caused by a type of TB resistant to the “first choice” of drugs for treatment. As such, the best strategy to stop the epidemic of tuberculosis (TB) is to isolate and treat infectious patients before they spread the disease.
The most common and transmissible form of TB is pulmonary TB. The diagnosis of pulmonary TB in Peru and most of the rest of the world relies on asking patients with clinical signs or those who have been exposed to the disease to provide two samples of sputum – the body fluid from the upper airways. Lab technicians then process the sample and look under a microscope for the presence of the infecting agent (Mycobacterium tuberculosis). In areas of highest incidence, like the laboratories of big hospitals or referral centers, this tedious task causes a bottleneck in both the diagnosis and treatment processes. This bottleneck is exacerbated by the need to continually test patient sputum samples and track the reduction of infecting agents as they go through treatment
In order to reduce or eliminate this bottleneck, a more efficient methodology for diagnosis and treatment is required.
Population Impacted
The TB epidemic took the lives of approximately 2 million people and 9 million people got infected worldwide in 2004 alone. In Latin America, Peru, Haiti and Bolivia are the countries most impacted by TB according to the last report from the World Health Organization (2006). It has been calculated that every untreated person with active TB will infect ten to twenty people throughout the duration of the disease.
Proposed Solution
Our proposal is to automate detection of TB in samples of sputum by acquiring and processing digital images of the microscopic photos of the samples. We propose to automate the microscopic search for the infecting agent by creating an annotated library, computer algorithms and hardware to acquire and analyze the digital images of the samples. We expect that implementation will be possible with a low-cost and low-maintenance interface, a commercial digital camera and a personal computer. Therefore, it would represent a reasonable low cost upgrade for hospitals with high service demands and would be flexible enough to be used in emergency rooms and small health centers.
Beneficiaries
Infected patients: Early diagnosis will greatly increase the chances of a cure.
Patients infected that do not respond to the first choice of drugs: Ability to better monitor progress during treatment will increase the chances of cure
Laboratory staff in hospitals: Ability to process and validate more samples and treat more patients in a shorter amount of time
Physicians: Delivery of test results in shorter time frames will facilitate better treatment
General population: Will benefit by reducing the burden on the public health system and ending the vicious correlation between poverty and TB.
We believe there are other tropical diseases diagnosed by finding microscopic pathogens that may benefit from the knowledge and expertise developed for this project.
Project Team
Lead Engineers
Benjamin Castaneda, University of Rochester. castaned@ece.rochester.edu
Roberto Lavarello, University of Illinois at Urbana-Champaign. lavarell@uiuc.edu
Engineering Students
Two engineering students in the Pontificia Universidad Catolica del Peru (PUCP).
Communications, Funding
Fanny Casado, University of Rochester. fanny_casado@urmc.rochester.edu
Others
We collaborate with Dr. Luz Huaroto, chief of the Laboratory of Mycobacteria in the Hospital Nacional Dos de Mayo which is the referral center in Lima Peru.
Obstacles and How Addressed?
Understanding the needs of the doctors and technicians including:
- the specific parameters to be measured
- the reality of clinical laboratories dealing with high volumes of samples, limited funding and training, and specific regulations regarding health worker safety, validation of procedures and patient’s privacy. We have had extensive conversations with doctors and technicians in the course of defining the project, collecting images, and presenting advancements for comments and feedback.
Project Status
Our efforts are moving slower than expected due to limited funding. Even though we have extended our search for funding locally and internationally, we have only secured 10% of the total funds needed. Currently, 70% of our funding is local.
We have succeeded in developing a preliminary annotated database of 450 images. We are currently developing preliminary algorithms for sample analysis which are targeted for completion by September, 2008. For validation purposes, we expect to collect around 1,000 images and then produce a final version of the software.
If the software developed shows promise, the next stage will be to develop hardware to automate not only the microscopy search for the pathogens, but also the image acquisition. Once that is done we will arrange for clinical trials of the methodology.
Assistance Sought
We have secured funding for the preliminary database and algorithms. We are seeking funding for the remaining tasks. We welcome interactions between the engineers and students in Peru with engineers and scientists around the world working on related topics. We anticipate that we will need assistance protecting our intellectual property rights, developing a business plan under a paradigm of social entrepreneurship and developing clinical trials for devices.