Breast cancer has high incidences and mortality rates in women worldwide.Malignancy could be detected manually by experienced pathologists based on Hematoxylin and Eosin (H&E) stained images. However, it is time-consuming and experience-dependent, making early diagnosis a big challenge. In this paper, a methodology for breast cancer classification based on histopathological images with deep learning was described. A residual learning-based convolutional neural network named myResNet-34 was designed for malignancy-and-benign classification. In addition, an algorithm automatically generating the target image for stain normalization was proposed, which eliminated the bias caused by manual selection of the reference image. Elastic distortion was introduced and combined with affine transformation for data augmentation considering the characteristics of the H&E images. Experiments were conducted on BreakHis dataset with the proposed framework. Promising results were achieved with an average classification accuracy of around 91% on image-level classification. Results indicated that both our data augmentation and stain normalization effectively improved the classification accuracy by 2-3%.

Cuffless method for blood pressure measurement is an important methods for continuous health status monitoring. A pulse wave is a periodic time-series signal that reflects a non-linear, non-stationary change in the pulse signal over time. Traditional ways of pulse wave based blood pressure assessment rely on feature extraction from pulse signals, which are usually signal quality dependent and lack of consistence among studies. In this paper, a method of blood pressure measurement of using continuous pulse waveform and long-term and short-term memory network is proposed, which avoids the process of manually extracting waveform features. Experiments were performed with both pulse wave signals and the arterial blood pressure signals form the MIMIC database. Empirical mode decomposition was applied for signal preprocessing, and the time series of the pulse wave was analyzed to establish a Long Short-Term Memory neural network for blood pressure assessment. An average prediction accuracy of 83.2% was achieved.