TB-v1: Software toolbox environment for the detection of TB bacilli in auramine staining

Tuberculosis and other mycobacteriosis are serious illnesses which control is based primarily in early diagnosis. In addition to clinical suspicion, a fundamental support in diagnosis is the staining of clinical samples with gender-specific colors. However, these techniques are not very sensitive so many times we must await the results of culture, which can take up to 2 months. We have developed a software called TB-v1 to help you through the application of techniques of digital image processing in specific staining of mycobacteria, providing specificity rates of around 98% for a sensitivity of 94% which justifies the conclusion that such system is susceptible to clinical applicability. The analysis techniques are based on the extraction of invariant descriptors from the shape of objects in conjunction with color information. The decisions taken by the program is based on Bayesian inference techniques. TB-v1 is a program that analyzes the images that have been captured by the camera attached to a fluorescence microscope and automatically indicates the presence or absence of bacilli in the image. 1. Background and state of the art Tuberculosis is a communicable disease, predictable and curable, so a key point in its control is rapid diagnosis. The slow growth of mycobacteria makes the diagnosis based on culture may be delayed up to two months. That is why methods based on staining are essential in this type of pathogen. The special structure rich in mycolic acids of mycobacteria makes specific stains exist for these organisms. Among the specific staining for mycobacteria, auramine stain is currently the most widely used due to its higher sensitivity and ease of observation. For observation using UV microscopy an objective of X20 is used. The estimated time to rule out as negative auramine stain is about 2 minutes. Tuberculosis is a disease that annually affects about 10 million people of which about 3 million die from the disease directly. In Madrid (Spain) the incidence of the disease has gone from as high as 40 cases per 100,000 inhabitants in the mid-nineties to the current estimate of 23 cases of tuberculosis per 100,000 inhabitants and a mortality of 1,5 cases (1999). Despite the large decrease in the incidence of tuberculosis in Madrid, it remains one of the highest rates of tuberculosis in Europe. There is no doubt that early diagnosis is the first weapon to control a communicable disease like tuberculosis. The increase in the number of people with inmnunosupresion has led to a rise in recent years several cases of mycobacterial tuberculosis which early diagnosis is also essential. Since the major clinically important mycobacteria (M. Manuel G. Forero Vargas y Gabriel Cristóbal Instituto de Optica (CSIC), Serrano 121, 28006 Madrid (Spain) Email: [email protected] tuberculosis complex, M. avium complex, M. kansassii) slow to grow in culture media from 1 to 8 weeks, the specific staining of mycobacteria have always been an essential diagnostic method in any mycobacteriology laboratory. Since more than a century ago Robert Koch described the first method of staining of mycobacteria, many techniques have been used for that purpose. From the point of view of the microscopy, two types of stains, non-fluorescent and those that are. Among the non-fluorescent staining, the stain most often used is, without doubt, the Ziehl-Neelsen. This staining technique used for coloring the basic fuchsin, the sulfuric acid to 25% for bleaching, and the methylene blue counterstain. Mycobacteria are shown colored red while the background is a bluish color. For viewing you must use the immersion objective (1000 ×) and 300 microscopic fields to be observed, equivalent to three lengths of a theoretical sample button 10x20 mm blade. In general, the time needed to rule out as negative preparation using this technique is about 10 minutes.