Is there a regression layer for Convolutional neural network toolbox?

Is there a regression layer for Convolutional neural network toolbox? I am trying to use "Convolutional Neural Networks" (CNN) to do a regression analysis for ultrasound images to calculate parameters which can have predictive value. The application for regression layer includes all parameters which are in the real domain and not just integers. Can I use CNN for that? 

Answer 1

[MathWorks Support Team]

Yes. The regression layer for Convolutional Neural Networks is appeared in R2017a Neural Network Toolbox. By defining regressionLayer at the last layer of the deep neural networks, the network computes regression models. Note that the last layer in classification is classificationLayer. 

In R2016b or previous versions, the regression layer is not supported. Please refer to the Release Notes for more information on recent changes to this toolbox:

https://www.mathworks.com/help/deeplearning/release-notes.html

[Michael Abboud]

This functionality is not supported in the R2017a release. However, you can still use an image datastore to get the image file names, then just create a table with the file names in the first column and your regressors in the other columns, and feed it into trainNetwork.