Development of an Automated Method for Classification of Masses in Mammogram.
スポンサーリンク
概要
- 論文の詳細を見る
We have developed an automated method for classification of masses in mammogram. The probability of malignancy was determined based on eight features extracted from mass shapes and analysis of mass spicules. The scheme consists of the following five steps: (1) Extraction of breast region, (2) Mass edge extraction, (3) Calculation of malignancy based on spicule patterns, (4) Extracting eight features from mass shapes and density, and (5) Classification by using artificial neural network (ANN). 599 cases including 17 masses with biopsy-proven were employed to estimate the performance of classification by the ANN. The scheme achieved a sensitivity of 82% at a specificity of 80%, and the area under the ROC curve (Az value) was 0.87. We conclude that our method is effective for discriminating mammographic masses.
- 一般社団法人 日本生体医工学会の論文
一般社団法人 日本生体医工学会 | 論文
- Evaluating the Workload Reduction of Automatic Vital Data Transmission
- Mental Fatigue Measurement Based on the Changes in Flicker Perception Threshold using Consumer Mobile Devices
- A Theoretical Study on a Computational Algorithm for Human Posture Estimation Based on Motion Capture of a Small Number of Markers
- Latest microscopic technique. From the principle to application. Development of three-dimensional microscopic system for operation of mini-pig fertilized eggs.
- A Study of the Automation of Cytodiagnosis (The Second Report):Cytophotometry Using Photomicroscanner and a Trial of Data-processing