SHIRC: A simultaneous sparsity model for histopathological image representation and classification

Abstract

Automated classification of histopathological images is an important research problem in medical imaging. Digital histopathology exhibits two principally distinct characteristics: 1) invariably histopathological images are multi-channel (color) with key geometric information spread across the color channels instead of being captured by luminance alone, and 2) the richness of geometric structures in such tissue imagery makes feature extraction for classification very demanding. Inspired by recent work in the use of sparsity for single channel image classification, we propose a new simultaneous Sparsity model for multi-channel Histopathological Image Representation and Classification (SHIRC). Essentially, we represent a multi-channel histopathological image as a sparse linear combination of training examples under suitable channel-wise constraints and classification is performed by solving a newly formulated simultaneous sparsitybased optimization problem. Experiments on two challenging real-world image databases: 1) provided by pathologists of the Animal Diagnostics Lab (ADL) at Pennsylvania State University, and 2) histopathological images corresponding to intraductal breast lesions [1], reveal the merits of the proposed SHIRC model over state of the art alternatives.

Classifying Multichannel UWB SAR Imagery via Tensor Sparsity Learning Techniques

“…Multi-channel SRC has been investigated before in medical images [16], [18]. In these papers, one dictionary for each channel is formed from training data with locations of all channels of one training point being the same in all dictionaries.…”

“…Similar to the SHIRC model proposed in [16], [18], we can impose one constraint on active elements of tensor x as follows: x also has few nonzero tubes as in SRC-CC; however, elements in one active tube are not necessarily the same. In other words, the locations of nonzero coefficients of training samples in the linear combination exhibit a oneto-one correspondence across channels.…”

IP on AP: Exploring Image Processing on the Automata Processor

“…We evaluate our results with two datasets: the ADL dataset [42], and the vehicle dataset using two different feature extraction and representation methods. We compare our results with SHIRC [42].…”

“…For every dataset, a threshold of T=10% reported the best accuracy. Table 3.1 shows a confusion matrix using our method at a 10% threshold in comparison with the simultaneous sparsity model for histopathological image representation and classification (SHIRC), which is a sparsity model that learns a dictionary for RGB color channels [42]. There are three confusion matrices for each of the organ datasets.…”

“…Windows 7 system equipped with Intel Core i7°2600 3.4-GHz processor and 8 GB RAM [42]. We compute the classification run-time in each method for one image and see a significant speedup.…”