![]() | Only 14 pages are availabe for public view |
Abstract In this thesis, three different approaches are proposed for increasing the automated cell morphology equipment performance. The first is based on the traditional medicalCAD system approach, the second is based on deep learning approach, and the third is basedon fusion of two previous approaches. The traditional pathological CAD system is developed in multi-dimensions through neutrosophic sets and GPU. A novel deep learning architecture is proposed to classifiy cells.The experimental results demonstrate that the proposed techniques arepromising with low complexity, adaptive and more robustness. This provides the basis for automatic medical diagnosis and further processing of medical images |