Cross-modality neural network transform for semi-automatic medical image annotation
View Patent ↗A cross-modality neural network transform for semi-automatic medical image annotation is provided. In various embodiments, an input medical image is mapped to a first vector in a text vector space. The first vector corresponds to the features of the medical image. A set of predetermined vectors is searched for a closest one of the predetermined vectors to the first vector. From the closest one of the predetermined vectors, one or more keywords is determined describing the input medical image.
1. A system comprising:
a trained convolutional network operative to receive as input a medical image and output an image feature vector corresponding to the medical image, the image feature vector having a first dimension;
a trained neural network operatively connected to the trained convolutional network to receive as input the feature vector and transform the image feature vector into a predicted document vector, the predicted document vector representing vectorization of a predicted text report, the predicted document vector having a second dimension;
a data store comprising a plurality of predetermined document vectors in a text vector space, the text vector space including the predicted document vector, the plurality of predetermined document vectors being determined by a trained learning system configured to receive a plurality of text reports and output a predetermined document vector for each text report, the predetermined document vectors having the second dimension, the data store operatively connected to the trained neural network; and
a computing node comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor of the computing node to cause the processor to perform a method comprising:
determining a closest vector of the predetermined document vectors to the predicted document vector,
determining, via concept extraction, one or more keywords describing the medical image from the closest vector, the one or more keywords comprising a semantic descriptor of the medical image, and
assigning the one or more keywords to the medical image.
2. The system of claim 1 , wherein the neural network is a feedforward neural network.
3. The system of claim 1 , wherein the closest vector of the predetermined document vectors is determined by distance within the text vector space of the predetermined document vectors to the first vector.
4. The system of claim 1 , wherein the closest vector of the predetermined document vectors is determined by Mahalanobis distance within the text vector space of the predetermined document vectors to the first vector.