Medical Information Retrieval through Mobile Devices

The proposed work deals with the development of a methodology for the classification of different stages of burnt images of human using image processing techniques. The burnt images of human at the leg part with different deepness of injuries are considered as dataset. The dataset includes the five classes representing phases of injuries, 10 images in each phase amounting to total of 50 images. The work mainly focuses on segmenting region into two parts, namely homogenous regions (background) and burnt part. For extracting the burnt region shearlet transform segmentation technique is applied. The shearlet transform is a multistate directional transform with a greater ability to localize distributed discontinuities such as edge region of wounded part. The shearlet transform found to be very efficient for segregating the injured part which further helped for the classification of burn stages. Further, features are extracted for wound region with texture and color features. The wavelet transform is adopted for texture extraction and color moments namely mean and standard deviation are extracted. The features are trained with the Support Vector Machine (SVM) classifier. The classification results are reported. The work helps for the dermatologist for disease diagnosis. Thus, the proposed system has the potential to facilitate information access and increase quality of patient care in clinical environment by making essential information available to the appropriate person at the appropriate time.