Patch Based Convolutional Neural Network for Whole Slide Tissue Image Classification

From statwiki
Revision as of 18:51, 15 November 2021 by Alivochk (talk | contribs) (EM-based method with CNN)
Jump to: navigation, search

Presented by

Cassandra Wong, Anastasiia Livochka, Maryam Yalsavar, David Evans


Despite the fact that CNN are well-known for their success in image classification, it is computationally impossible to use them for cancer classification. This problem is due to high-resolution images that cancer classification is dealing with. As a result, this paper argues that using a patch level CNN can outperform an image level based one and considers two main challenges in patch level classification – aggregation of patch-level classification results and existence of non-discriminative patches. For dealing with these challenges, training a decision fusion model and an Expectation-Maximization (EM) based method for locating the discriminative patches are suggested respectively. At the end the authors proved their claims and findings by testing their model to the classification of glioma and non-small-cell lung carcinoma cases.

Previous Work

EM-based method with CNN

The high-resolution image is modelled as a bag and patches extracted from it are instances that form a specific bag. The ground truth labels are provided for the bag only, so we model the labels of an instance (discriminative or not) as a hidden binary variable. Hidden binary variables are estimated by the Expectation-Maximization algorithm. A summary of the proposed approach can be found in Fig.2:

fig2 model architecture.jpeg
Figure 2: Top: A CNN is trained on patches and EM-based method iteratively eliminates non-discriminative patches.
Bottom: An image-level decision fusion model is trained on histograms of patch-level predictions to predict the image-level label

In this paper <math>X = {X_1, X_n}<\math> denotes dataset containing Nbags. A bag Xi= {Xi, 1, Xi, 2, Xi, Ni} consists of Nipathes (instances) and Xi, j = <xi,j, yj> denotes j-th instance and it’s label in i-th bag. We assume bags are i.i.d. (independent identically distributed), Xand associated hidden labels H are generated by the following model: