Thermogram assisted detection and analysis of Ductal Carcinoma In Situ (DCIS)

The main aim of this work is to develop a computer assisted tool to support automated detection and analysis of Ductal Carcinoma In Situ (DCIS) using Digital Infrared Thermal Image (DITI). Medical image inspection is a necessary practice in most of the disease recognition procedures. In the proposed work, a hybrid approach combining the Fuzzy-Entropy based Active Contour (FEAC) is employed to segment the abnormal regions in the chosen DITI. The performance of the proposed approach is then validated against the Fuzzy-Entropy based Region Growing (FERG) and Fuzzy-Entropy based Chan-Vese (FECV) approaches. The proposed work is implemented as follows; i) Assessment of symmetry in breast, ii) Cropping and region based analysis of the normal/abnormal breast region, and iii) Computation of the Haralick texture characteristic, iv) Adaptive Neuro Fuzzy Inference System (ANFIS) based illness classification. All the experimental works are implemented using the Matlab software. During this investigation, 50 DITI images (25 normal and 25 abnormal) are considered and the results of this study shows that, FEAC + ANFIS approach offers classification accuracy for normal, abnormal and intermediate DITI images of 97.14%, 98.52% and 94.38% respectively.