Bosom disease (BC) is one of the most continuous kinds of malignancy that grown-up females experience the ill effects of around the world. Numerous BC patients face irreversible conditions and even passing because of late determination and treatment. Thusly, early BC analysis frameworks dependent on obsessive bosom symbolism have been sought after in ongoing years. In this paper, we acquaint an end-with end model dependent on completely convolutional network (FCN) and bidirectional long momentary memory (Bi-LSTM) for BC discovery. FCN is utilized as an encoder for elevated level element extraction. Yield of the FCN is gone to a one-dimensional arrangement by the smooth layer and took care of into the Bi-LSTM's information. This technique guarantees that high-goal pictures are utilized as immediate contribution to the model. We directed our investigations on the Break his information base, which is openly accessible at http://web.inf.ufpr.br/vri/bosom malignant growth information base. So as to assess the execution of the proposed strategy, the exactness metric was utilized by considering the five-overlap cross validation procedure. Execution of the proposed technique was discovered to be in a way that is better than already announced outcomes.