• - Chronic Obsrtuctive Pulmonary Disease (COPD) is a progressive lung diseasis which includes Emphysema. Emphysema is an illness which is charaterised by increasing breathlessness mostly due to the lung damage caused by the lung exposure to smoke.

    - The first objective of this project is to find out if a patient suffers from Emphysema by using the texture analyses of the CT scans taken from the patient's lung

    - The second objective here is to find out what part of lung has been damaged and how severe the damage is to the lung.

    - To achieve these two objectives, initially CT scan images are analysed with texture analysis techniques. Two dimensional texture analysis methods are used to extract features from slices of the lung CT scans[1,2,3,4]. Since lung CT scans are three dimesional images, three dimensional texture feature extractions are then used for emphysema diagnosis [5,6].


    • COPDSample

    - In the above figure, The top row shows some lungs with two different kinds of Emphysema [2].

    - In the second row, the areas of the lung damaged by emphysema is detected. The last remaining rows show a probabilistic map of the lung tissues for being healthy or being damaged by two kinds of emphysema [2].

    [1]- Dharmagunawardhana, Chathurika, Mahmoodi, Sasan, Bennett, Michael and Niranjan, Mahesan (2014) Gaussian Markov random field based improved texture descriptor for image segmentation. Image and Vision Computing, 32 (11), 884-895. (doi:10.1016/j.imavis.2014.07.002).

    [2]- Dharmagunawardhana, Chathurika, Mahmoodi, Sasan, Bennett, Michael and Niranjan, Mahesan (2014) Quantitative analysis of pulmonary emphysema using isotropic Gaussian Markov random fields. 9th International Conference on Computer Vision Theory and Applications, Portugal. 05 - 08 Jan 2014. pp. 44-53.

    [3]- Dharmagunawardhana, Chathurika, Mahmoodi, Sasan, Bennett, Michael and Niranjan, Mahesan (2014) An Inhomogeneous Bayesian Texture Model for Spatially Varying Parameter Estimation. 3rd International Conference on Pattern Recognition Applications and Methods, France. pp. 139-146 .

    [4]- Dharmagunaw, C., Mahmoodi, S., Bennett, M. and Niranjan, M. (2016) Rotation invariant texture descriptors based on Gaussian Markov random fields for classification. Pattern Recognition Letters, 69, 15-21. (doi:10.1016/j.patrec.2015.10.006).

    [5]- Al Makady, Yasseen, Mahmoodi, Sasan, Conway, Joy and Bennett, Michael (2018) Volumetric texture analysis based on three-dimensional Gaussian Markov random fields for COPD detection. Nixon, M., Mahmoodi, S. and Zwiggelaar, R. (eds.) In Annual Conference on Medical Image Understanding and Analysis: MIUA 2018: Medical Image Understanding and Analysis. vol. 894, Springer. pp. 153-164 . (doi:10.1007/978-3-319-95921-4_16).

    [6]- Al Makady, Yasseen, Mahmoodi, Sasan and Bennett, Michael (2019) Gaussian Markov random fields-based features for volumetric texture segmentation. In Proceedings of IEEE International Conference on Multimedia Information Processing and Retrieval (MIPR). IEEE. 4 pp .