An automated deep learning model for liver tumor and skin lesion segmentation

Artificial Intelligence plays a vital role in automatically identifying different diseases from a given set of medical images. Medical images require accurate organ segmentation to detect abnormalities in different organs of a body using diverse medical image formats, which allows the specialists to timely diagnosis a disease. In this context here we developed a ResUNet which is an advanced hybrid deep learning model which performs automatic segmentation of liver tumors and skin lesions through multiple convolutional neural network (CNN) layers. The performance of this model is evaluated through segmentation metrics like Dice Score, Jaccard Index, Volumetric Overlapping Error and Accuracy. The liver tumor segmentation displayed a volumetric overlapping error of 33.10% and a Jaccard Index of 63.12% while the skin lesion confirmed a volumetric overlapping error of 11.00% and a Jaccard Index of 88.99%. Additionally, the Liver tumor and skin lesion segmentation achieved a DSC of 75.21%, 94.04% and Accuracy of 98.55%, 96.95% respectively.