DDS4 - Dataset to evaluate Tuberculosis diagnosis by automatic sputum smear microscopy

DDS4 - Dataset to evaluate Tuberculosis diagnosis by automatic sputum smear microscopyThe number of acid-fast bacilli found in the sputum smear of the patient indicates the infectiousness of the patient. The grading scales for reporting the results from reading a smear are comprised of 5 classes: negative, scanty, 1+, 2+ and 3+ (Rieder et al, 2007).This dataset was generated to evaluate a proposal of Tb diagnosis by automatic sputum smear microscopy. It is composed of 250 sets of Extended Depth Focus (EDF) images of microscropy fields: 50 sets corresponding to the negative diagnostic class composed of 164 images; 50 sets corresponding to a scanty diagnosis composed of 164 images; 50 sets corresponding to a 1+ diagnosis composed of 164 images; 50 sets corresponding to a 2+ diagnosis composed of 82 images; and 50 sets corresponding to a 3+ diagnosis composed of 32 images. This dataset was generated as a testing set used to evaluate automatic smear microscopy for Tb diagnosis. There is no annotation in these images.The depth resolution of the images is 24 bits (true color images) and the dimensions are 1388 × 1040 pixels. The EDF images are digital images obtained by the fusion of a stack of 11 sputum smear images of the same microscopy field acquired at different focal depths.The folders are individually stored as tar.gz archives, for a total size of 49 GBH.L. Rieder, A. Van Deun, K. Man Kam, S. Jae Kim, T.M. Chonde, A. Trebucq, et al.Priorities for Tuberculosis Bacteriology Services in Low-Income Countries, vol. 2 (2007)