1. How can AI be utilized in breast cancer detection using ultrasound imaging?
AI can be utilized in breast cancer detection using ultrasound imaging by developing software that can perform tasks previously only achievable with human intelligence. AI has been increasingly used in ultrasonography and has shown to be a powerful tool for delivering trustworthy detection with greater accuracy and efficiency while minimizing the effort of physicians. Various methods have been proposed, including groupings, Neural Networks (NN), and scalar feature selection methods. For instance, a CNN model design was proposed for the classification of breast ultrasound images, achieving a precision of 97% and an AUC of 98%. Additionally, a CNN model was used to extract features from a convolution layer for classifying breast cancer ultrasound images, achieving an overall accuracy of 99.1% and an overall recall of 97%.
read more
2. What is the image size and format of dataset group A?
Dataset group A consists of 780 images with a mean size of 500 pixels by 500 pixels. The images are in PNG format. The original images are accompanied by ground truth images and are categorized into three classes: benign, malignant, and normal. The dataset was collected in 2018 and includes breast ultrasound images of females aged 25-75. The total number of patients is 600, all of whom are women. The images in group A were used in the study and are in PNG format.
read more
3. What classification model was used in the study?
In the study, a Convolutional Neural Network (CNN) model was implemented to classify breast ultrasound images into two groups. The CNN model extracted elevated features from the images and employed them to detect relevant texture features for classification. Different transfer learning methods, such as VGG-16, VGG-19, Squeeze Net, and DarkNet-53, were used to enhance the feature extraction process. The CNN model achieved 96% accuracy with data A and 100% accuracy with data B.
read more
4. What CNN model was used for breast cancer ultrasound image analysis?
The CNN model used for breast cancer ultrasound image analysis consists of four layers. It was designed to classify cancerous, non-cancerous, and normal images. The model achieved the best classification accuracy for data B from data A, making it a viable solution for Iraq's challenging clinical environment. By assisting radiologists in the diagnosis process and reducing the time spent on normal breast images, this model aims to increase the number of patients receiving medical care.
read more