Computer-aided classification of mammographic masses using visually sensitive image features.

PURPOSE: To develop a new computer-aided diagnosis (CAD) scheme that computes visually sensitive image features routinely used by radiologists to develop a machine learning classifier and distinguish between the malignant and benign breast masses detected from digital mammograms. METHODS: An image dataset including 301 breast masses was retrospectively selected. From each segmented mass region, we computed image features that mimic five categories of visually sensitive features routinely used by radiologists in reading mammograms. We then selected five optimal features in the five feature categories and applied logistic regression models for classification. A new CAD interface was also designed to show lesion segmentation, computed feature values and classification score. RESULTS: Areas under ROC curves (AUC) were 0.786±0.026 and 0.758±0.027 when to classify mass regions depicting on two view images, respectively. By fusing classification scores computed from two regions, AUC increased to 0.806±0.025. CONCLUSION: This study demonstrated a new approach to develop CAD scheme based on 5 visually sensitive image features. Combining with a "visual aid" interface, CAD results may be much more easily explainable to the observers and increase their confidence to consider CAD generated classification results than using other conventional CAD approaches, which involve many complicated and visually insensitive texture features.

Feasibility of Electroacoustic Tomography: A Simulation Study.

The feasibility of electroacoustic tomography (EAT) was investigated for in situ monitoring the electric field distribution in soft tissue. EAT exploits the phenomenon that the amplitude of acoustic emission generated by an electric field is proportional to the electrical energy deposition in tissue. After detecting these acoustic waves with ultrasound transducers, an image of the electric field distribution can be reconstructed in real-time. In our computer simulations, the electric field distribution in soft tissue was generated by solving general partial differential equations (PDEs) using finite element analysis (FEA). The electric field distributions were converted into initial pressure distributions, and the propagation of the induced acoustic waves was simulated using K-Wave simulation. A circular array of 128 ultrasound transducers was placed around the target to detect the acoustic waves, and a time reversal reconstruction algorithm was used to reconstruct the EAT image. A different number of electrodes set at different distances with different voltage inputs on the electrodes were performed to simulate different electric field distributions during electroporation. It was found that the electrical energy deposition in reconstructed EAT imaging is decreased as the distance of the electrodes increases. We also have investigated the sensitivity of the EAT imaging with different voltage inputs. The minimal voltage we can detect with EAT is 970 V at the pulsewidth of 180 ns. The results of this study demonstrated that EAT is a feasible technique for monitoring the electric field distribution and guiding the electrotherapy in future clinical practice.