Wuhan University

Juan Liu

Jing Feng

Zhuoyu Li

Yuqi Chen

Zhiqun Zuo

Jiasheng Liu

Chunbing Hua

Na Zhang

The present disclosure provides a multi-modal method for classifying a thyroid nodule based on ultrasound (US) and infrared thermal (IRT) images. Based on ultrasound and infrared thermal images and in combination with a multi-modal learning method, the present disclosure provides an adaptive multi-modal hybrid (AmmH) model which is composed of three parts: an intra-modal hybrid encoder (HIME), an adaptive cross-modal encoder (ACME), and a multilayer perceptron (MLP) head. The HIME is capable of modeling a global feature while extracting a local feature. The ACME is capable of customizing personalized modality-weights according to different cases and performing information interaction and fusion of inter-modal features. The MLP head classifies a fused feature obtained. The method enables the AmmH model to automatically classify a thyroid nodule of a subject based on ultrasound and infrared thermal images of the subject, providing a doctor with an objective and accurate classification result to assist diagnosis.

Multi-modal method for classifying thyroid nodule based on ultrasound and infrared thermal images

Provided is a method and system for predicting tumor mutation burden (TMB) in triple negative breast cancer (TNBC) based on nuclear scores and histopathological whole slide images (WSIs). The method includes the following steps: first, screening the histopathological WSIs of TNBC; calculating a TMB value of each patient according to gene mutation of each patient with TNBC, and dividing the TMB values into two groups with high and low TMB according to a set threshold; dividing the histopathological WSIs of TNBC into patches of a set size; screening a certain number of patches with high nuclear scores according to a nuclear score function; then building a convolutional neural network (CNN) classification model, and stochastically initializing parameters in the CNN classification model; and finally, putting the screened patches into the built CNN classification model for training, so as to automatically predict high or low TMB with the histopathological WSIs of TNBC.

Method and system for predicting tumor mutation burden (TMB) in triple negative breast cancer (TNBC) based on nuclear scores and histopathological whole slide images (WSIs)

Provided is an end-to-end attention pooling-based classification method for histopathological images obtaining a better classification effect for small number of samples by S1, cutting the histopathology image into patches of a specified size, removing the patches with too much background area and packaging the remaining patches into a bag; S2, training a deep learning network by taking the bag obtained in S1 as an input using a standard multi-instance learning method; S3, scoring all the patches by using the trained deep learning network, and selecting m patches with highest and lowest scores for each whole slide image to form a new bag; S4, building a deep learning network including an attention pooling module, and training the network by using the new bag obtained in S3; and S5, after the histopathology image to be classified is processed in S1 and S3, performing classification by using the model obtained in S4.

End-to-end attention pooling-based classification method for histopathology images

An automatic abnormal cell recognition method, the method including: 1) scanning a slide using a digital pathological scanner and obtaining a cytological slide image; 2) obtaining a set of centroid coordinates of all nuclei that is denoted as CentroidOfNucleus by automatically localizing nuclei of all cells in the cytological slide image using a feature fusion based localizing method; 3) obtaining a set of cell square region of interest (ROI) images that are denoted as ROI_images; 4) grouping all cell images in the ROI_images into different groups based on sampling without replacement, where each group contains ROW&#xd7;COLUMN cell images with preset ROW and COLUMN parameters; obtaining a set of splice images; and 5) classifying all cell images in the splice image simultaneously by using the splice image as an input of a trained deep neural network; and recognizing cells classified as abnormal categories.

Automatic abnormal cell recognition method based on image splicing

An automatic classification method of whole slide images (WSIs) for cervical tissue pathology based on confidence coefficient selection. The automatic classification method includes steps: S1: dividing the WSIs for the cervical tissue pathology into small pieces having set size, gathering the small pieces of each WSI into a packet, and removing blank pieces in the packets; S2: building a deep CNN model; S3: training the deep CNN for designated rounds; S4: performing sequential arrangement and connection to obtain feature vectors of WSIs by using the trained deep CNN as the feature extractor; S5: training a support vector machine classifier; and S6: processing the WSIs for the cervical tissue pathology, to be classified, through step S1 and step S4 to obtain the feature vectors of the images, and inputting the feature vectors into the trained support vector machine classifier to realize classification.

Growing community of inventors

Wuhan, China

Juan Liu