Mammography is one of the effective methods to identify breast cancer and it plays a crucial role in reducing death toll by detecting cancer at an early stage. However, some factors such as high volume screening, the complexity of X-ray images particularly when radiologist conducts reading while he/she is tired can lead to false diagnosis results.
The presence of breast microcalcifications during the screening shows that the patient has an early stage of breast cancer. Breast calcifications appear like scattered spots in the X-ray image that range from 0.1 to 1.0 mm in size. The accurate detection highly depends on the expertise of the radiologist or CADx (Computer-Aided Diagnosis) that serves as the second opinion for radiologists to identify breast cancer with high accuracy.
Our technology Mammo Cancer AI that based on a Computer-Aided-Diagnosis (CADx) has been developing on purpose to improve the detection accuracy and assist the physicians and radiologists in their overloaded work. The CADx system assists junior radiologists by increasing their sensitivity from 62% to 90% and experienced radiologists by increasing their sensitivity from 77% to 90%. With assistance of the system, the malignancy of this lesion is recognized by its shape, asymmetry and contour.
The Deep Convolutional Neural Network has been trained 1 000 000 digital mammograms so it could classify benign or malignant tumors from 454 X 454 pixels mammographic windows. Our Neural Network strategy uses a number of dense layers which creates higher accuracy in predicting the malignancy of unseen windows, providing information about location detected lesions.
Mammo Cancer AI is a web-based software works right in a web browser that makes it easy to use anytime, anywhere. The Web solution uses big data analytics and artificial intelligence for early and accurate breast cancer detection. The tool is handy for radiologists when they are tired, it assists to detect invisible areas and it does not require expert skills.